From speechsc-bounces@ietf.org Tue Mar 1 06:30:51 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA13942 for ; Tue, 1 Mar 2005 06:30:50 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D65bS-0000Wv-Lx for speechsc-web-archive@ietf.org; Tue, 01 Mar 2005 06:31:51 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D65Yl-0003iv-9Z; Tue, 01 Mar 2005 06:29:03 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D65Yk-0003iq-KN for speechsc@megatron.ietf.org; Tue, 01 Mar 2005 06:29:02 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA13809 for ; Tue, 1 Mar 2005 06:28:59 -0500 (EST) Received: from tidos.tid.es ([193.145.240.2] helo=correo.tid.es) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D65Zf-0000UE-K0 for speechsc@ietf.org; Tue, 01 Mar 2005 06:30:00 -0500 Received: from correo.tid.es (filvit [192.168.48.202]) by tid.hi.inet (iPlanet Messaging Server 5.2 Patch 2 (built Jul 14 2004)) with ESMTP id <0ICO00MWM7NOUH@tid.hi.inet> for speechsc@ietf.org; Tue, 01 Mar 2005 12:23:48 +0100 (MET) Received: from tid.es (valldigna.hi.inet [10.95.12.129]) by tid.hi.inet (iPlanet Messaging Server 5.2 Patch 2 (built Jul 14 2004)) with ESMTPA id <0ICO001BB7NKVA@tid.hi.inet> for speechsc@ietf.org; Tue, 01 Mar 2005 12:23:47 +0100 (MET) Date: Tue, 01 Mar 2005 12:22:02 +0100 From: =?ISO-8859-1?Q?Jes=FAs_Bernat_Vercher?= To: speechsc@ietf.org Message-id: <4224505A.1080501@tid.es> Organization: =?iso-8859-1?Q?Telef=F3nica_I=2BD?= MIME-version: 1.0 Content-type: text/plain; format=flowed; charset=ISO-8859-1 Content-transfer-encoding: 7BIT X-Accept-Language: es-es, es User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; es-ES; rv:1.6) Gecko/20040113 X-imss-version: 2.7 X-imss-result: Passed X-imss-scores: Clean:99.90000 C:22 M:1 S:5 R:5 X-imss-settings: Baseline:1 C:1 M:1 S:1 R:1 (0.0000 0.0000) X-Spam-Score: 0.0 (/) X-Scan-Signature: d17f825e43c9aed4fd65b7edddddec89 Content-Transfer-Encoding: 7BIT Subject: [Speechsc] load-lexicon/define-lexicon X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: cf4fa59384e76e63313391b70cd0dd25 Content-Transfer-Encoding: 7BIT In the draft-ietf-speechsc-mrcpv2-06.txt it appears define-lexicon as a method (page 53), but in the state machine and in other places of the document (page 33, 34, ...) it appears a method called load-lexicon. It seams to me that both names refer to the same thing and there should be just one. Right? Jesus Bernat (Telefonica I+D) _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Tue Mar 1 08:52:30 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA25811 for ; Tue, 1 Mar 2005 08:52:30 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D67oa-0003gT-8I for speechsc-web-archive@ietf.org; Tue, 01 Mar 2005 08:53:33 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D67lK-00040p-M6; Tue, 01 Mar 2005 08:50:10 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D67lH-00040d-0q for speechsc@megatron.ietf.org; Tue, 01 Mar 2005 08:50:07 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA25500 for ; Tue, 1 Mar 2005 08:50:03 -0500 (EST) Received: from salvelinus.brooktrout.com ([204.176.205.6]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D67mD-0003dY-U6 for speechsc@ietf.org; Tue, 01 Mar 2005 08:51:06 -0500 Received: from nhmail2.needham.brooktrout.com (nhmail2.brooktrout.com [204.176.205.242]) by salvelinus.brooktrout.com (8.12.5/8.12.5) with ESMTP id j21DiT67016464 for ; Tue, 1 Mar 2005 08:44:29 -0500 (EST) Received: by nhmail2.brooktrout.com with Internet Mail Service (5.5.2653.19) id ; Tue, 1 Mar 2005 08:40:35 -0500 Message-ID: From: Eric Burger To: "IETF SPEECHSC (E-mail)" Date: Tue, 1 Mar 2005 08:40:35 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" X-Spam-Score: 0.0 (/) X-Scan-Signature: 6d62ab47271805379d7172ee693a45db Subject: [Speechsc] Out of the Office X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 7bac9cb154eb5790ae3b2913587a40de Rather than spamming the list with "Out of Office Autoreply", I'll just spam the list with, "I'm out of touch." I'm sure many of you think I'm out of touch normally, but, really, I probably will not have e-mail access until the Monday of IETF. _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From zhoiz@msn.com Fri Mar 4 16:30:17 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA05592; Fri, 4 Mar 2005 16:30:17 -0500 (EST) Received: from ip-229.net-82-216-227.roubaix.rev.numericable.fr ([82.216.227.229] ident=dchs) by ietf-mx.ietf.org with smtp (Exim 4.33) id 1D7KOv-0006Xo-CW; Fri, 04 Mar 2005 16:32:02 -0500 Received: from [96.152.188.64] (helo=..dearriba.com) by smtp2.cistron.nl with esmtp ( 3.35 #1 ()) id 306LFL-0038PT-14 Message-ID: <12102093144732.R37415@.noc..gr> Sender: freeradius-devel-zhoiz@msn.com X-Mailman-Version: 2.0.1 Date: Sat, 05 Mar 2005 14:31:44 +0100 From: "Adan Evans" To: speechsc@ietf.org, speechsc-admin@ietf.org, speechsc-request@ietf.org, speechsc-web-archive@ietf.org, spirits-archive@ietf.org, ssion@ietf.org, ssm@ietf.org, ssm-admin@ietf.org, ssm-archive@ietf.org, ssm-request@ietf.org Subject: SU-per Hu^ge 0ffers Speechsc X-Spam-Score: 10.8 (++++++++++) X-Spam-Flag: YES X-Scan-Signature: d6b246023072368de71562c0ab503126 Biggest stores of on'line pharmacies. Over than 150 major meds Highest qualities assured on our online generic meds at the most competitive prices. Vi'sit us today's! http://859fs.com/_fd5977142df59d3662baa654773c6a8e/ This is 1 -time mailing. N0-re m0val are re'qui-red 1DtGPhwKWkJ0nV5Ku5Ac2fmvxpsRmTwF3p47q From speechsc-bounces@ietf.org Sun Mar 6 07:33:50 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA03024 for ; Sun, 6 Mar 2005 07:33:50 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D7uzC-0000vA-In for speechsc-web-archive@ietf.org; Sun, 06 Mar 2005 07:35:55 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D7uvl-0006H2-1G; Sun, 06 Mar 2005 07:32:21 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D7uvj-0006Gw-9y for speechsc@megatron.ietf.org; Sun, 06 Mar 2005 07:32:19 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA02951 for ; Sun, 6 Mar 2005 07:32:17 -0500 (EST) Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D7uxh-0000tA-Ic for speechsc@ietf.org; Sun, 06 Mar 2005 07:34:22 -0500 Received: from ams-core-1.cisco.com (144.254.224.150) by ams-iport-1.cisco.com with ESMTP; 06 Mar 2005 13:52:13 +0100 X-BrightmailFiltered: true X-Brightmail-Tracker: AAAAAA== Received: from imail.cisco.com (imail.cisco.com [128.107.200.91]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id j26CW5lu005291 for ; Sun, 6 Mar 2005 13:32:06 +0100 (MET) Received: from [10.32.245.155] (stealth-10-32-245-155.cisco.com [10.32.245.155]) by imail.cisco.com (8.12.11/8.12.10) with SMTP id j26CPolL010962 for ; Sun, 6 Mar 2005 04:25:50 -0800 Mime-Version: 1.0 (Apple Message framework v619.2) Content-Transfer-Encoding: quoted-printable Message-Id: <4437449815adc2cfa8c287c19466dbc1@cisco.com> Content-Type: text/plain; charset=WINDOWS-1252; format=flowed To: "'speechsc@ietf.org'" From: David R Oran Date: Sun, 6 Mar 2005 07:32:02 -0500 X-Mailer: Apple Mail (2.619.2) IIM-SIG: v:"1.1"; h:"imail.cisco.com"; d:"cisco.com"; z:"home"; m:"krs"; t:"1110111951.684832"; x:"432200"; a:"rsa-sha1"; b:"nofws:8534"; e:"Iw=="; n:"sQYarK2E51MdcTiUqeif3F7cWdxIfoCiXhdfb9vD5ee/j0jXL15gbFxF2p" "XIweAblu0N6XAgK7k+wrbr7bQDJaCDqOmzqpRUBjIRQAXQ7NzadpmR3pUL6wxaRUtW+c43sl9jC" "50Qg1sXHpPjt8Y+Y16ioyQAQAdSunM4YhevURc="; s:"CRWbTPuanQ0hwvxJhcVWSpFH+/y7yGHx5J8pkB9jY/rLcLq2dSM6Q/T9EXi5jCqgZEZBtLFm" "aN7tBdxieQ3H+z4aWsjVtn2wDhpsF32rd+d6oJRcIDpnfRJdNT5VRvlkbzZJVxrfNDopxqRqZrd" "CC2mnbNfBXZwRvpHpww3GlUE="; c:"From: David R Oran "; c:"Subject: Advance Material for SI/SV Security Discussion at SpeechSC " "WG meeting Tuesday"; c:"Date: Sun, 6 Mar 2005 07:32:02 -0500" IIM-VERIFY: s:"y"; v:"y"; r:"60"; h:"imail.cisco.com"; c:"message from imail.cisco.com verified; " X-Spam-Score: 0.0 (/) X-Scan-Signature: a743e34ab8eb08259de9a7307caed594 Content-Transfer-Encoding: quoted-printable Subject: [Speechsc] Advance Material for SI/SV Security Discussion at SpeechSC WG meeting Tuesday X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: ff0adf256e4dd459cc25215cfa732ac1 Content-Transfer-Encoding: quoted-printable Here is some advance reading to frame the discussion. The IESG has a=20 "discuss" on the SpeechSC requirements document over the issue of=20 security and privacy concerns about providing speaker verification and=20= identification services. This note contains some points I gleaned from reading the document=20 cited by the IESG (and which is also already cited in the security=20 considerations of section of the Requirements draft). I'd appreciate any thoughts you have on this to inform the discussion.=20= Please post as response to the list so we have a record. If any of you wish to spend the $22 to otain the report, that wold be=20 very helpful. It's probably worth the money because there's a lot of=20 good stuff in there (despite my disagreement with the conclusions=20 related to SI and SV). See draft-ietf-speechsc-reqts-06.txt for the=20 exact citation. Thanks, Dave (co-chair SPEECHSC and draft-ietf-speechsc-reqts-06.txt=20 author) ------------------------------------------------ I thought it prudent to read the entire NRC report which Steve Bellovin=20= used as a major justification for his "discuss". So, I swallowed my=20 annoyance at having to spend the $22 to download the PDF (otherwise you=20= can only read piecemeal and can't search). Here are what I believe are=20= the relevant quotes from the report: "Recommendation: If biometrics are used to uniquely identify license=20 holders and to prevent duplicate issuance, care must be taken to=20 prevent exploitation of the resulting centralized database and any=20 samples gathered. (6.3)" - Box Es.1 on P.21. "Biometrics has the obvious advantage of authenticating the human, not=20= just the presented token or password. Common biometrics in use today=20 verify fingerprints, retinas, irises, and faces, among other things. =20 Downsides to biometrics include the fact that not all people can use=20 all systems, making a backup authentication method necessary (and=20 consequently increasing vulnerability); the fact that revocation is not=20= possible for current systems (the saying goes that most individuals=20 =93have only two thumbs=94); and that remote enrollment of a biometric=20= measure (sending one=92s fingerprint or iris scan over the Internet, for=20= example) may defeat the purpose and is easily compromised. " followed=20 by "Recommendation: Biometric technologies should not be used to=20 authenticate users via remote authentication servers because of the=20 potential for large-scale privacy and security compromises in the event=20= of a successful attack (either internal or external) against such=20 servers. The use of biometrics for local authentication=97for example,=20= to control access to a private key on a smart card=97is a more=20 appropriate type of use for biometrics. (5.2) " - N.B. This seems to be=20= the key point of contention. - External attacks against a speaker verification system can be foiled=20= by well-understood security means, and these are in fact employed by=20 speechsc - TLS encryption of the control channel, SRTP encryption of=20 the media channel, and authentication/authorization of all elements in=20= the chain from the speaker to the server holding the voiceprints. - Internal attacks can be simply the stealing of the database or the=20 compromise of the server system, including its keying material. There=20 is the pragmatic issue of a server with voiceprints being like a=20 /etc/passwd file, so it is of course prudent to store the data=20 encrypted so it can't be stolen en masse. But speechsc is addressing a=20= protocol standard, so it isn't clear why we need to say more than we do=20= about that aspect, or more than is said about server security in the=20 Radius specs. Server compromise can leak voiceprint information. At=20 issue is whether it is the use of the voiceprints for authentication,=20 or the mere storage of the voiceprints which causes the security=20 concerns. For internal attacks, it appears to be the latter which is=20 the privacy issue. - Speaker verification differs from other biometric authentication=20 techniques because it is easier to make it resilient to replay or=20 impersonation attacks. Unlike static biometrics like retina, face=20 geometry, fingerprints, etc., speaker verification can (and is) done=20 via challenge response protocols. The subject is asked to say a number=20= of words or a phrase chosen by the verification system, and the result=20= matched against the vocal characteristics captured in the voiceprint.=20 Possession of a stolen voiceprint does not by itself enable=20 impersonation, as would a fingerprint or retina scan (c.f "The Sixth=20 Day", "Mission Impossible") etc. - voiceprints of varying quality can be obtained by anything which can=20= record your voice. A voiceprint used for identification/verification=20 needs to be protected, but reality dictates that not all recording=20 devices will be super-secure, hence requiring those holding voiceprints=20= to be more secure than those holding speech recordings seems=20 questionable. The "security first" tradeoff would say well, then all=20 voice recording things need to be as secure as the servers holding=20 voiceprints. Are we prepared to go that far - revisit all the specs=20 like RTSP, SIP, XCON, etc. to ensure that they cannot be used to make=20 recordings that can be turned offline into voiceprints? - an accurate voiceprint is in practice a biometric of the same=20 confidence as a fingerprint, retina scan, etc., with the exception that=20= it can't be used unless you can be induced to speak enough to allow a=20 match (sort of like the old call trace scene in the movies where the=20 bad/good guy hangs up before the listeners have enough time/info to=20 trace the call). Superficially it is more subject to surreptitious=20 acquisition and mishandling than other biometrics, but in reality I=20 don't believe it is given how easy it is to lift fingerprints from=20 identifying objects, get DNA samples from personal items, etc. - speaker identification can be done by anything which can hear you=20 speak. If you speak in a non-secure session, there are obvious problems=20= (including surreptitious recording, as noted above). If you speak in a=20= secure session with authentication (by means other than speaker=20 verification, rendering the identification issues moot) it seems there=20= is an implicit agreement that your identity can be ascertained by the=20 participants, by your voice or other means with or without the aid of=20 computational services. If you speak in a secure session with explicit=20= anonymity, there is a higher degree of expectation that if the context=20= is one in which you are not identifiable by other means (e.g.=20 historical, correlation with prior sessions that were not anonymous,=20 etc.) that no voiceprint exists, or if it does exist it is of limited=20 utility (e.g. it can identify you as "the one who's been interrupting=20 on the last three calls"). While the existence of speaker=20 identification in this situation does limit privacy somewhat, it is not=20= out of line with common expectations of privacy in live communication=20 situations, or after the fact for that matter. Those who are truly=20 concerned that indirect methods of identification (such as speaker=20 identification) be thwarted are the people who do things like use voice=20= distorting devices, rendering the identification and verification=20 systems impotent. - Many other authentication methods also have privacy weaknesses of=20 various sorts. As the report points out (cf. p.67): "Private secrets are items of information that are so intimately=20 associated with an individual or with events in the (human) individual=92s life that no other person (or few others) would be expected to know about them. The use of private secrets for authentication causes several=20 problems. People resist the use of private secrets for authentication on the=20 grounds that they are private and should not have to be revealed to third=20 parties (even to third parties who wish to authenticate us). Private secrets are rarely completely private.23 This leads to another problem: Any item of information that is used as a private secret to authenticate an=20 individual will typically be shared with all the people and organizations that=20 want to authenticate the individual (technical measures exist that could prevent sharing this, but they are not widely used). Each party who=20 authenticates the individual therefore comes to know the information that is supposed to be a private secret, and thus the information becomes less private=20 and less secret as time goes by. Not only do these problems compromise the=20= individual=92s privacy, but they also gradually destroy the usefulness of the information as an=20 authenticator. When the information=92s value as an authenticator has been substantially degraded, another piece of (private) information will have to be chosen as a new basis for the private-secret authentication=20 process. If this sequence of events is repeated enough times, many of the individual=92s private secrets will have been revealed to large numbers = of other parties. The individual=92s privacy will have been put at=20 considerable risk, as will the ability of other parties to authenticate the=20 individual. Social Security numbers are an excellent case study in this phenomenon, as a recent incident involving the Princeton University admissions=20 office illustrates. Both Princeton University and Yale University use Social=20 Security numbers as student identifiers. Yale=92s admissions office used the Social Security number as an authentication key to allow students to=20 access the status of their applications for admission via a Web browser. A member of the Princeton University admissions office staff discovered this and apparently used Social Security numbers obtained from the records of applicants to Princeton in order to access the Yale=20 admissions Web site and learn about Yale=92s admissions decisions." _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Sun Mar 6 15:56:37 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA24911 for ; Sun, 6 Mar 2005 15:56:37 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D82pq-0005OI-Nb for speechsc-web-archive@ietf.org; Sun, 06 Mar 2005 15:58:46 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D82mG-0002f6-00; Sun, 06 Mar 2005 15:55:04 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D82mF-0002eu-GW for speechsc@megatron.ietf.org; Sun, 06 Mar 2005 15:55:03 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA24839 for ; Sun, 6 Mar 2005 15:55:01 -0500 (EST) Received: from salvelinus.brooktrout.com ([204.176.205.6]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D82oI-0005Mu-Nt for speechsc@ietf.org; Sun, 06 Mar 2005 15:57:10 -0500 Received: from nhmail2.needham.brooktrout.com (nhmail2.brooktrout.com [204.176.205.242]) by salvelinus.brooktrout.com (8.12.5/8.12.5) with ESMTP id j26KpspB026324 for ; Sun, 6 Mar 2005 15:51:54 -0500 (EST) Received: by nhmail2.brooktrout.com with Internet Mail Service (5.5.2653.19) id ; Sun, 6 Mar 2005 15:47:58 -0500 Message-ID: From: Eric Burger To: speechsc@ietf.org Date: Sun, 6 Mar 2005 15:47:55 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" X-Spam-Score: 0.0 (/) X-Scan-Signature: 7aefe408d50e9c7c47615841cb314bed Subject: [Speechsc] For our remote guests X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 68c8cc8a64a9d0402e43b8eee9fc4199 mp3 streaming: http://videolab.uoregon.edu/events/ietf/ietf62.html jabber chat: http://www.xmpp.org/ietf-chat.html _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Tue Mar 8 14:19:27 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA03802 for ; Tue, 8 Mar 2005 14:19:27 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D8kHI-00057h-8o for speechsc-web-archive@ietf.org; Tue, 08 Mar 2005 14:22:00 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D8kEm-0000e9-MR; Tue, 08 Mar 2005 14:19:24 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D8kEk-0000e3-HZ for speechsc@megatron.ietf.org; Tue, 08 Mar 2005 14:19:22 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA03789 for ; Tue, 8 Mar 2005 14:19:20 -0500 (EST) Received: from web14121.mail.yahoo.com ([66.163.171.112]) by ietf-mx.ietf.org with smtp (Exim 4.33) id 1D8kHB-00057U-OR for speechsc@ietf.org; Tue, 08 Mar 2005 14:21:54 -0500 Received: (qmail 53792 invoked by uid 60001); 8 Mar 2005 19:19:20 -0000 Comment: DomainKeys? See http://antispam.yahoo.com/domainkeys DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; b=afz42qbBTBt7+q9K8pEoid9M4tEjfhX1qA4pwz29bdYOr6mc3ppzDpxo3NPe0VZKelv4l7rRb9dO+24Rk3oTW8WkzWntChOXn1svFJn+YpQvdHdoTWx5KYJhUfogwUQMFagX9f+eED6c5FiSs6NQTuOKH5stRO3jGnGYH4gJw+8= ; Message-ID: <20050308191920.53790.qmail@web14121.mail.yahoo.com> Received: from [63.193.248.104] by web14121.mail.yahoo.com via HTTP; Tue, 08 Mar 2005 11:19:20 PST Date: Tue, 8 Mar 2005 11:19:20 -0800 (PST) From: Dan Burnett Subject: Re: [Speechsc] Advance Material for SI/SV Security Discussion at SpeechSC WG meeting Tuesday To: David R Oran , "'speechsc@ietf.org'" In-Reply-To: <4437449815adc2cfa8c287c19466dbc1@cisco.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Spam-Score: 3.0 (+++) X-Scan-Signature: 65215b440f7ab00ca9514de4a7a89926 Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by ietf.org id OAA03789 X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 3.0 (+++) X-Scan-Signature: 3b8eea209b62bd15620865bc4fbef8cd Content-Transfer-Encoding: quoted-printable Good analysis. I have one comment embedded below which is a minor elucidation of a point you already make. -- dan --- David R Oran wrote: > Here is some advance reading to frame the > discussion. The IESG has a=20 > "discuss" on the SpeechSC requirements document over > the issue of=20 > security and privacy concerns about providing > speaker verification and=20 > identification services. >=20 > This note contains some points I gleaned from > reading the document=20 > cited by the IESG (and which is also already cited > in the security=20 > considerations of section of the Requirements > draft). >=20 > I'd appreciate any thoughts you have on this to > inform the discussion.=20 > Please post as response to the list so we have a > record. >=20 > If any of you wish to spend the $22 to otain the > report, that wold be=20 > very helpful. It's probably worth the money because > there's a lot of=20 > good stuff in there (despite my disagreement with > the conclusions=20 > related to SI and SV). See > draft-ietf-speechsc-reqts-06.txt for the=20 > exact citation. >=20 > Thanks, Dave (co-chair SPEECHSC and > draft-ietf-speechsc-reqts-06.txt=20 > author) >=20 > ------------------------------------------------ >=20 > I thought it prudent to read the entire NRC report > which Steve Bellovin=20 > used as a major justification for his "discuss". So, > I swallowed my=20 > annoyance at having to spend the $22 to download the > PDF (otherwise you=20 > can only read piecemeal and can't search). Here are > what I believe are=20 > the relevant quotes from the report: >=20 > "Recommendation: If biometrics are used to uniquely > identify license=20 > holders and to prevent duplicate issuance, care must > be taken to=20 > prevent exploitation of the resulting centralized > database and any=20 > samples gathered. (6.3)" - Box Es.1 on P.21. >=20 > "Biometrics has the obvious advantage of > authenticating the human, not=20 > just the presented token or password. Common > biometrics in use today=20 > verify fingerprints, retinas, irises, and faces, > among other things. =20 > Downsides to biometrics include the fact that not > all people can use=20 > all systems, making a backup authentication method > necessary (and=20 > consequently increasing vulnerability); the fact > that revocation is not=20 > possible for current systems (the saying goes that > most individuals=20 > =93have only two thumbs=94); and that remote enrollment > of a biometric=20 > measure (sending one=92s fingerprint or iris scan over > the Internet, for=20 > example) may defeat the purpose and is easily > compromised. " followed=20 > by >=20 > "Recommendation: Biometric technologies should not > be used to=20 > authenticate users via remote authentication servers > because of the=20 > potential for large-scale privacy and security > compromises in the event=20 > of a successful attack (either internal or external) > against such=20 > servers. The use of biometrics for local > authentication=97for example,=20 > to control access to a private key on a smart > card=97is a more=20 > appropriate type of use for biometrics. (5.2) " - > N.B. This seems to be=20 > the key point of contention. >=20 >=20 > - External attacks against a speaker verification > system can be foiled=20 > by well-understood security means, and these are in > fact employed by=20 > speechsc - TLS encryption of the control channel, > SRTP encryption of=20 > the media channel, and authentication/authorization > of all elements in=20 > the chain from the speaker to the server holding the > voiceprints. >=20 > - Internal attacks can be simply the stealing of the > database or the=20 > compromise of the server system, including its > keying material. There=20 > is the pragmatic issue of a server with voiceprints > being like a=20 > /etc/passwd file, so it is of course prudent to > store the data=20 > encrypted so it can't be stolen en masse. But > speechsc is addressing a=20 > protocol standard, so it isn't clear why we need to > say more than we do=20 > about that aspect, or more than is said about server > security in the=20 > Radius specs. Server compromise can leak voiceprint > information. At=20 > issue is whether it is the use of the voiceprints > for authentication,=20 > or the mere storage of the voiceprints which causes > the security=20 > concerns. For internal attacks, it appears to be the > latter which is=20 > the privacy issue. Voiceprints typically consist of voice characteristics compiled from multiple user utterances. The original utterances cannot be derived from the contents, ensuring that access to a voiceprint alone does not enable the direct creation of impostor utterances. I believe you allude to this in the next paragraph. >=20 > - Speaker verification differs from other biometric > authentication=20 > techniques because it is easier to make it resilient > to replay or=20 > impersonation attacks. Unlike static biometrics like > retina, face=20 > geometry, fingerprints, etc., speaker verification > can (and is) done=20 > via challenge response protocols. The subject is > asked to say a number=20 > of words or a phrase chosen by the verification > system, and the result=20 > matched against the vocal characteristics captured > in the voiceprint.=20 > Possession of a stolen voiceprint does not by itself > enable=20 > impersonation, as would a fingerprint or retina scan > (c.f "The Sixth=20 > Day", "Mission Impossible") etc. >=20 > - voiceprints of varying quality can be obtained by > anything which can=20 > record your voice. A voiceprint used for > identification/verification=20 > needs to be protected, but reality dictates that not > all recording=20 > devices will be super-secure, hence requiring those > holding voiceprints=20 > to be more secure than those holding speech > recordings seems=20 > questionable. The "security first" tradeoff would > say well, then all=20 > voice recording things need to be as secure as the > servers holding=20 > voiceprints. Are we prepared to go that far - > revisit all the specs=20 > like RTSP, SIP, XCON, etc. to ensure that they > cannot be used to make=20 > recordings that can be turned offline into > voiceprints? >=20 > - an accurate voiceprint is in practice a biometric > of the same=20 > confidence as a fingerprint, retina scan, etc., with > the exception that=20 > it can't be used unless you can be induced to speak > enough to allow a=20 > match (sort of like the old call trace scene in the > movies where the=20 > bad/good guy hangs up before the listeners have > enough time/info to=20 > trace the call). Superficially it is more subject to > surreptitious=20 > acquisition and mishandling than other biometrics, > but in reality I=20 > don't believe it is given how easy it is to lift > fingerprints from=20 > identifying objects, get DNA samples from personal > items, etc. >=20 >=20 >=20 > - speaker identification can be done by anything > which can hear you=20 > speak. If you speak in a non-secure session, there > are obvious problems=20 > (including surreptitious recording, as noted above). > If you speak in a=20 > secure session with authentication (by means other > than speaker=20 > verification, rendering the identification issues > moot) it seems there=20 > is an implicit agreement that your identity can be > ascertained by the participants, by your voice or > other means with or without the aid of computational > services. If you speak in a secure session with > explicit anonymity, there is a higher degree of > expectation that if the context is one in which you > are not identifiable by other means (e.g. historical,=20 > correlation with prior sessions that were not anonymous,=20 > etc.) that no voiceprint exists, or if it does exist it is of limited=20 > utility (e.g. it can identify you as "the one who's been interrupting=20 > on the last three calls"). While the existence of speaker=20 > identification in this situation does limit privacy somewhat, it is not=20 > out of line with common expectations of privacy in live communication=20 > situations, or after the fact for that matter. Those who are truly=20 > concerned that indirect methods of identification (such as speaker=20 > identification) be thwarted are the people who do things like use voice=20 > distorting devices, rendering the identification and verification=20 > systems impotent. > - Many other authentication methods also have privacy weaknesses of=20 > various sorts. As the report points out (cf. p.67): > "Private secrets are items of information that are so intimately=20 > associated with an individual or with events in the (human) individual=92s life > that no other person (or few others) would be expected to know about > them. The use of private secrets for authentication causes several problems. > People resist the use of private secrets for authentication on the grounds > that they are private and should not have to be revealed to third parties > (even to third parties who wish to authenticate us). > Private secrets are rarely completely private.23 This leads to another problem: Any item of > information that is used as a private secret to authenticate an individual > will typically be shared with all the people and organizations that want to > authenticate the individual (technical measures exist that could prevent > sharing this, but they are not widely used). Each party who authenticates > the individual therefore comes to know the information that is supposed > to be a private secret, and thus the information becomes less private and > less secret as time goes by. Not only do these problems compromise the individual=92s privacy, but > they also gradually destroy the usefulness of the information as an authenticator. > When the information=92s value as an authenticator has been > substantially degraded, another piece of (private) information will have > to be chosen as a new basis for the private-secret authentication process. > If this sequence of events is repeated enough times, many of the > individual=92s private secrets will have been revealed to large numbers of > other parties. The individual=92s privacy will have been put at considerable > risk, as will the ability of other parties to authenticate the individual. > Social Security numbers are an excellent case study in this phenomenon, > as a recent incident involving the Princeton University admissions office > illustrates. Both Princeton University and Yale University use Social Security > numbers as student identifiers. Yale=92s admissions office used the > Social Security number as an authentication key to allow students to access > the status of their applications for admission via a Web browser. A > member of the Princeton University admissions office staff discovered > this and apparently used Social Security numbers obtained from the > records of applicants to Princeton in order to access the Yale admissions > Web site and learn about Yale=92s admissions decisions." =09 =09 __________________________________=20 Celebrate Yahoo!'s 10th Birthday!=20 Yahoo! Netrospective: 100 Moments of the Web=20 http://birthday.yahoo.com/netrospective/ _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Tue Mar 8 15:03:16 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA10080 for ; Tue, 8 Mar 2005 15:03:16 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D8kxh-0006y7-UW for speechsc-web-archive@ietf.org; Tue, 08 Mar 2005 15:05:50 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D8ks7-0007u0-L3; Tue, 08 Mar 2005 15:00:03 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D8kab-0002fz-Q3 for speechsc@megatron.ietf.org; Tue, 08 Mar 2005 14:41:58 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA05891 for ; Tue, 8 Mar 2005 14:41:55 -0500 (EST) Received: from sj-iport-3-in.cisco.com ([171.71.176.72] helo=sj-iport-3.cisco.com) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D8kd0-0005cl-Hr for speechsc@ietf.org; Tue, 08 Mar 2005 14:44:29 -0500 Received: from sj-core-2.cisco.com (171.71.177.254) by sj-iport-3.cisco.com with ESMTP; 08 Mar 2005 12:58:49 +0000 X-IronPort-AV: i="3.90,147,1107734400"; d="pdf'?scan'208"; a="232727739:sNHT150871464" Received: from mira-sjc5-d.cisco.com (IDENT:mirapoint@mira-sjc5-d.cisco.com [171.71.163.28]) by sj-core-2.cisco.com (8.12.10/8.12.6) with ESMTP id j28Jfeq8018515 for ; Tue, 8 Mar 2005 11:41:41 -0800 (PST) Received: from [127.0.0.1] (ssh-sjc-1.cisco.com [171.68.225.134]) by mira-sjc5-d.cisco.com (MOS 3.4.6-GR) with ESMTP id AIJ76057; Tue, 8 Mar 2005 11:41:04 -0800 (PST) Mime-Version: 1.0 (Apple Message framework v619.2) To: "'speechsc@ietf.org'" Message-Id: Content-Type: multipart/mixed; boundary=Apple-Mail-1-1007160049 From: David R Oran Date: Tue, 8 Mar 2005 14:41:02 -0500 X-Mailer: Apple Mail (2.619.2) X-Spam-Score: 1.1 (+) X-Scan-Signature: f25395c61ce97c691771e8bd63d587fc X-Mailman-Approved-At: Tue, 08 Mar 2005 15:00:02 -0500 Subject: [Speechsc] Chair slides for Speechsc WG meeting today X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 1.1 (+) X-Scan-Signature: 2bc3a2301894f2d19c9ee8f5860dbac8 --Apple-Mail-1-1007160049 Content-Type: text/plain; charset=US-ASCII; format=flowed Content-Transfer-Encoding: 7bit For our remote participants... --Apple-Mail-1-1007160049 Content-Type: application/pdf; x-unix-mode=0644; name="speechsc-62.ppt.pdf" Content-Disposition: inline; filename=speechsc-62.ppt.pdf Content-Transfer-Encoding: base64 JVBERi0xLjMKJcTl8uXrp/Og0MTGCjIgMCBvYmoKPDwgL0xlbmd0aCAxIDAgUiAvRmlsdGVyIC9G bGF0ZURlY29kZSA+PgpzdHJlYW0KeNq1V1tzWzUQftev0KOZoVuttLq9tQ3pwAOXjs3wHNyUBppA Ygp/n5V8jqQj+RiXoZN04s737Wq1l0/rR/lGPkrj+BeUUii9VtKSkk+38if5IJ9fHVDuDxLzz2Gf qNa7THt25Il3F/pQ+Yd9PEoFmo7/Tx+MEypZKhvl/l6+2kkbMsp/MCip+d/uXj5/jaA4it07uQlf yN2vDU2jBiItJupG9rgO4CjOrjbf9rj3KRKGRYJvOtioCC7oYv7U4zoCKj/hYrPvcRtAeyr273s8 KvDoZA1fLHBSDqKu5+vOnkwE46p/lXDR4B4hKLfEG3uriK8fCm4zfr0Tbz5rbRXZU8V1XXScWc/V FWu3R+TyhBr9w4CTBmNKdd8OuCfwXq41DyKXJ9bqftPjWnnAev71gBNnrzbfrqsu6oDga3Vf9/YG PRi7jE+0uDFga3WfDfZWQwzD/UTBHbfv+nAg8fm+2t81uMi44fal9fyTM21+OlwgRQvOFPy2t7c8 varW52bAnQKLpT/+GPDI96/5+73HnbZgYumPDwNuQpv/uwG3CAZLfxwGPChopuvLAY8OXFydfvSs buFMfbzD1v93fX8EZcDh+vkBeX70en+ENP0V/3Gw9wGw5mfb90fk/mnU5+WkLvJCdRFn1EWuqov2 wS/URRzVBT/x9F7bxJnTOVOZnP7G6HiyIvXqlrMjWiYi8vMQCnvzQw5xQdGk0ogXynXukQXFOO4S aimDFwoWUFfK1ejFca1sQ/l69OIp8mtUKdvRS/AaFuf8e87Fhe/JkHMznWuMQHROxvFFmVJRmZxQ SxDNTJ5e9OSjMmJMr0ZcTP3Ch2FdYl2ZGfvRh0nKpgtDjj7IaKhBb16NPqxGqISPowvLh3h77iqO uwcr45fRh0eCetfb0YUPAag/5MIpEseKckcovslFVZ2aKfWSYjRQFF1NjzG2vKCTSjM1R/hzj2Pa AaOY8Y8D7i0oKvZPPa6VBh/nUKYktjiPBZpiP8SnuUys0kdcjP6NNqCr/YseJ94yWKXr/USHc0ej X4+feMuI1T6/gqLB0yur3RJv7S1vSfyKzPhvEy5m3KFLKjbjf/b2jpdUMifjExl3eYtaPd+jScq0 Uj/B5eNXzK+f74NJr9iMQ4/zgsz5K/2xH3DySV9X4wuR0hY198d9nRAr/z4q29W2mYPt1bk5EDwH U+sryR+jo/zxg3wvt5/j5VzIAfI67Fxu1Hbmvjqho567Nk7keTdbqiTl7X1m/HVCA1l8EIuP2xOM kHqz+DihozbtcMXF9yPB8Xc0hcXF0wkGX4Xw3FU8X8X64uPhP6wSx+KLTxVB7i5pjW8eNpEKMjSh NjaJ4EQ9IWJpFS/wTa8h6YsuUcEfBo2yNmngjL/oNYT4oTGh4PtRgzSoGt5dryFW5TVnxg+DBvE3 OdbIU/6zhlie8ZqpMT3O5FTPOPQa4qxJ3wRW43es4cqt+/fEm7iY4f9DAnLhWVbv05qO+WORgH8A zG+flAplbmRzdHJlYW0KZW5kb2JqCjEgMCBvYmoKMTEyMwplbmRvYmoKNCAwIG9iago8PCAvVHlw ZSAvUGFnZSAvUGFyZW50IDUgMCBSIC9SZXNvdXJjZXMgNiAwIFIgL0NvbnRlbnRzIDIgMCBSIC9N ZWRpYUJveApbIDAgMCA3OTIgNjEyIF0gPj4KZW5kb2JqCjYgMCBvYmoKPDwgL1Byb2NTZXQgWyAv UERGIC9UZXh0IF0gL0NvbG9yU3BhY2UgPDwgL0NzMSAzIDAgUiA+PiAvRm9udCA8PCAvRjEuMAo3 IDAgUiA+PiA+PgplbmRvYmoKOSAwIG9iago8PCAvTGVuZ3RoIDggMCBSIC9OIDMgL0FsdGVybmF0 ZSAvRGV2aWNlUkdCIC9GaWx0ZXIgL0ZsYXRlRGVjb2RlID4+CnN0cmVhbQp42n2ST0gUURzHv7Ml QqwFZSZS8E62B1cG7WAdjN31b8q2rGumCLLOvtkdnZ2d3sxuJR5CiC5B1jG6WNFJOoYHDx0CDxGC Yl0i6CgZBIKXkO03M7vuiNqDN+8zv/+/33tAXShtmnqAAXnDFsn+KLs7PsHqN1CHBgRBK61YZiSR GHaZbHFk7X2F5Jyb4eP1/10NghICElWAxqzH1xye9njA4fu2aRNPOqzk0hlik7hNpJIx4lfEZ7M+ nvZxhlsK8TLxTcUUFCeQIh4oKVkn5g6xbGQ0g+SXibsylpInJt/AU2cWXpn2ENB9BTj1uSabsIDl d8Cl1pos1AxcHANWOmuy3aQ7H6lp3VI7O1yRFIwCdT/K5d1WoP4FsP+8XP77ulzef0M5vgMfdaUo SpUZSdIXwOvDXY393OBCU5hzXwlRUDWd+0Z6vNrr14tH9SWrdBJ7M3FXF7BE9zB2Bhh6DLz8CVx9 D1z4ACQagNR1BB6hum3+wM0TK5gPhZbN2axDljtY2Dk6WYReCGexQt4s2lywQUNpb2NpXWeuqcUE t7go8Uw78nqx2u852kFujI7QSfMKqNzqrbA0k0n30N2gnXgjw3t6nXdBvKhqfYPOhdD+pIq+UY+l 85o9mPI40G3o8eEKwyjEb3sxsWPa0WQ1vlUa6a3KZ9K3EnS2kPzGbGHIsWki39BcLjXmsZSay8Xi FV7FOHRwaDDoa4AhiX5EEYYJgQJU0mhkoZGUuzaC2MLssZY6Ej5mpN8mn23X5x6K5O143UE0joW2 gwhM/ib/lrfkJfmt/GuxpRiqaRbElKasP/tDcZ3M1bgVbaUmL75CeSOk1ZElaf6gJ8tXqa861Vhs qUVycvAn8T1fl9yXKYxpN9Ksm6nk6iz6RnzZTpoe2a7NrzbXcm2dXpncDK7NH5pV4UhX/KCrw/81 O78//wfNsAFoCmVuZHN0cmVhbQplbmRvYmoKOCAwIG9iago3MDYKZW5kb2JqCjMgMCBvYmoKWyAv SUNDQmFzZWQgOSAwIFIgXQplbmRvYmoKMTEgMCBvYmoKPDwgL0xlbmd0aCAxMCAwIFIgL0ZpbHRl ciAvRmxhdGVEZWNvZGUgPj4Kc3RyZWFtCnjavX1Lsx03cuYev+LErOgIdRnvx+wsmXRwMY62mxPe 9Ia6IiV6RFJ9eWmZ/34yUQeFrExkyd0dlB5BKZAHBWR+yBeAxF9u/3b7yy1k+Gez1rpb8faWor09 vrn9x+3D7R+/++RuD59urv/96QFJU8md7A87nXn7P+zD9r+hj7/c7Obj/v/4HyEbi7+0qd0e3t++ fXVLtbfCH67am4d/X72//eMLt1kYxau3t2f1H26v/pOQeee3GL25kz678XZftxzb6OrZ/+HtpeBI oNlg82vWHGzbcvXHzx95u2+bs+Xebp498PZUN1/i8fufeHuzW3H5NodvTu3R5q35+X3Pfh9D20Ke /VtsN6S9uK3afG4nv082wvTr0Z56+/NX5t++qmxtTCvhZjY64GwB6Rpt9s6BeOoc/QfRHv0WwiHd H0R7iVspNw08zoF42pTuS97ubdnc/P5z0R6BexN8r5h0na9uK1O6L/jvgytbSOfxGdoewpamdP8g fp/81qqYnznaM8BXXxwuwvfL/P070m56ewD4Rp3/MQfKH9ZuXGxpy+Fof8N/n2D12imf16I92y25 Ax+/iPYG85/8+8jbs09baAc+fhbtoVL+vxPtyW3BHfj4JNqr3cjq+ka0t7zlpq5+V0C71Qv5lOxo ///K8VFt2LLTv18drB+v46Pi6p/t/1f8vtTNTf78ieOjAX6I9vmnu3a5/Q+1i7nQLjdVu/hSy0m7 mF27+L/y61y3mYuvA6c6Mf4JWglY5kArM/X2r128J1If4hYP4jtATxQBBgnfO0ie3sheYm2bT5Pm 9h+yn+LC5o9xmftaO5FUl/pqHd38vCBJCexRJyEr5uvztIKWcC0Klv75meBG83GzyR3koHXFNCyO Mk6SnyRPXaizfcEsD5a35Ulye7GQi/Vong6ad7KbGPsKPUg+yF5STFsql4PJsWwpkMH8UXZTPejq NGkeF9IFc3QarugEfJHThz6ITrwF/oPDdZA8/fkfej9/O0iOZW8kSHLprfCHTwknWXaI+AMiu9I5 6EwAnQhsP2jvc5gd4YqL4BUcBF84AVjN6szRfuPtlXXwiRFEdFp9b++r6DNvB4mXFGYH33OC0jZb 8yAwz94zggSGuUU3e3jHCWLbYg50jOZMUNIW7cUksgXbl/gnDCFA41LIJz7yHhKoJJulJMwgAEYm sM8qp4vFHqbcnz0dBGYn8OhgFX0MBRQaHQL9As6lFNbBE++gITLJGH9ieKsefLTqJ8Eb1kMFNoB6 1mdZEW+hTsC95ATg5ljUV4PgOSNowW8lT8C94u0lgB9BRPWCETjrLHqaB+BugsBDIBV0wKHP3/XP FLZhBB3UOqfB2S9bZnwkeHMOhuCavrCdQ3eRjPEHhjfw98NWsy4q50GYMckeBt6cTxk9ZlWWMMZz D98zwLlgzz1w7eNCtJvT4WLA6z/3IBgZYG0HFbFotyBoDkZnQwS/2PuLWcYU0XE9APedIGgYuOrr 0iWAbG4TcEKYKXffVJ9lqnFLRM8/csBl5HS8gCyGB3Ttf88RlyG8J5P4LDpobmtBjvEAXIFlRdEg 2FDAPyiNToIBrmTwyC5UpCsVGEk+8ZYDrlqQ1dUYQPec1tUjB1yFdXWlpiFOgDiAfOIXTtAc9mB0 Tjbkg9ctomvzE8QzpQQQa6aqS9tbFzc3Afcg2tFVI7N8LQhgafvTujOMoHX/Sh2BQy54JglDCXAI SddwHuCwRIM5CNq5BzEJ1HA+rsy+uRPEdkLDTfQAiKzKJ0wnAElQB4yLygeIVVpWCYwPoBzi1RgC DvIC0z5U7MFomL75CKsiX30ievR/JuB+EQQpXuMlgoKaeDFyDAkUVLxQ5D75REfI4Jai2/KFjvWp LPTCAZUE5vZCx/vszsbyNUdKjmdT9+GvCRbNdXLxN+KAknqUS+MA7jT1zO+glC4Tpt6aByM4CLjP hKndHNPsgftMAWAO/mcnMCsOYhgQMD4cPfAhxgAWbBJIIxgzDtLNHjhCIoYiMUyCN9zNdyCCQAge RSQBpp4SfGBKKYGHnBBm8xMyUHApy0EaEihkjJcGwR94DwDkWgij/plBLYMTbr3XZ4GJNE8JuFIq +AlHCN7yHnLcLCVgrDZoZ1OIk0AEK+DygJGZBHw91tB7MOosKmiUcvWJCrKIE3D/zt18B5qZiopD tmG+OPoJOO63NdAJ0VZ9BA6tqC2U0dzNRxuWGZ7OcQD5uXTR0UaGiwE4X3QxOwDiBkpD/7qrDTNc B8EX7ux0A5kuBtAzQ6SHT4IADWQ26qJ1HgxDqOtZ7A48GsgyBS09fISSviRdyB43Jg45vxf+OX4h XvSAiat0UgtMzhFckZp11dST/6ExRlLvOsGitWQW72UIAIN0kk8HWBJ4O0mCiTj4MMawmqUZDn7e or8Qdgafz5JJ/MoBhxkITz4hgowCcClJ1yvQUDeKaRGuFXArs7v4RIU144gpE6CvaKnzxTTRw3fh AnC4E5DTxSCbc7gVdSDujfTwXU9zqmNoyAcCyQeOuFbLSZrC37EgC5d1U+Zt2nOXGma7i2+9BMz0 r60/yeK/hYce0mlh8YyVdziGC1mAj++GLMQ0d//annt4K338fFLBPIMAPn44KVkZJRR/0nEiEMEs hk9Gn0Xw5x5eShc+bXd7a1YOGrjwoEYJJ19JDz1v0+8wwtp53J7MF+YUPHS7Bd2WgYPeeppjQpJ7 2LC4Y2XSNidntW72Yu2Cj55Peva/OOIyjiFf9JB9PVmTJ+mnQ9DILd7v4qbj0Qrmpovhw4pL6Lnt tNKc+oaZbDCng0Ck48HFdYH0wKWESCv3T5iV4giYmZzjlXA/PPn7hGTGHzeS6Rj4NCMYmorZqkHw A8NaBBcXtOwk4Ao0gRddMMk7CB65ow4KtlEC0cPw5OcYmKOOmSJML2iszh57IIziMSFumLsrTmZw feLk5CIh36MJwqj/xd1shHsmPbzkjnrCVDYheC4cdTDZZJavFn66RXs6CF4sEva2XvCpVsyF54k4 4WcHWEAY2I0eOCQbqK402SBDw9YAUBQvMp0O0UB0FC/cUw+l2zINs86m3C3RIBD5UWBCt0SD4LN0 1v1W5Rintw+YvVo24M+3bgQkn8xM6buiA86hMUxujSezp/T31OMg2GRKf09/KtKGkKHuuclB8Cfh jvt4YqTgU0gQdxEd9yAI2CdkTt8fgzTrlH049yDClr5dOCXF3f0KS9+vVv7hzbu+bXGS9cnd922j nxAeYgJZU0vwnqMlwyfshYp0OdqNKrgPwt8v8WLF5OZPc/wkXHVXT3iXyXjw/qigZTSQMD9rdLiW mrsfrSKhgrlLZQr6Z5nOT5u7wFrNMIkwNYuImhqo4Cs7AktuT41ONnBPPdbuRmvaDVz5cw+PwlO3 qP90uHkbcJt/EvzM/Sab3AkLwvmzuXYnWErbjHR+7IGZhhcP6q1v3qwGafZ8vu+umd4DxF3Nqaw2 Hk8oJrkkpqOOX0i6rMCTB68g6oADTz5dzzL4MGZplg4aJuPpupKxQM7EEhnpAwbwClrQlbiPsPB8 1hHnMZVKe3gps/XltDKfc8TFVvq2huYVeMyUJsKoF1/xLC33o23jx16EFCqeCEFB77QCCQF3+Dve 7wSftHMxQLAUNCaVYnda7j28EwdjcAzHJ4xY99HhisqzB+FoR3BJMN8yB8ncZAwXEpnFN7wHPK/h SA887YSbLrWSMbwWGXHkQ5KMGqolIR/EJ06ONvqHpIdfuZuMhyk86UHk1DFcoJ/4mamWjFv80a0J 9pQ57t4EdZAGt6YTHQPnQ+nnEJoOmII79HiOQOMDpswL/cSvwtO2+xGiAbnHxdkZ23WLAjnMiKeW J+SexNkZnEWZPXCC1g9DnGRhmK9ez8ISrniPq7wOuVZZD6+lr17woK74hJmuuOsbnoNAGG3bA6uV uIejbM89CKvu9sPeKuTAWW+7ElZ7KHWjsH8QjjLwOkZdFt1Zb+QTj4IAN37zhNxHmVvvmXFNVi7g HglB3DuZWo+7RV1PAnPre+p8IO4mfXW7x1VaDxGWHhH2R+Fr33Pvg+C9llo/EZx87divvRwEn+Xh GbsbMw1wuPvbLnS9y7hHH1azHL427kH4C9Dne157hQYzTs+4cDEGzAXkdNPQYPoB+lYv4FCQk17X ca5ifj/rOg7c7bJ7HpqwKgTJpU7EvZap87aFcrG2cduuuIk46ZAD6P2FQXKt1dO6EplOC5wknsYi Mw6IskXXceBOh5P6kB5725POGuS8u6etB8EX4S1DgNiaruP68Rjag/S3Wz8Cv0TUnlsH2BenS7Of fkm6EjT99AvVs9Ibhii2ehW098x4MqoSBH+6nFTUbeVP35eecnbFnXsQR3QiBJG1TMjJwy3Ah5h1 JedRR9mkm1WffOzZec2T6+dbgruAXAKfvhXdvfB4AtARVv8kkuuYv0+6hgFpnz/xIJLrtZzw8NPv l11PofDsuhATnpFsg1RCqR9GbeZof7OICUIhHYjkOzpoiAMgWGo3PIiVQ5o9/CBiggSqqR7TkQRg 8gPuoWiTwEuQCeE+Z8EOwYDrYxvpgWv5BP64pYP8xIMGPDhB+SBy6+h2zFkscus4TToLkVsPrIen RW7dkiF8XKXWcSt6EPxvAkXzdwaoBxQNv44VcZFH+A8WoP45eKutBfN3DOC4DzZusUU0CYsBPPG7 sCABb8tBe78VRAgieH4YjQwCfl8xFRgMqqxBwC9s4pHkGskn+I1S3GHxmGweBM/ZlTxUSM2ST7zi PUDgl/Bk9SCYl0pNJygWNVZlgyQ9FLzjg37XIPiFXesrCe9EkB74vcmCx9JCUhllCgafnhDwm6MV d4cb4QO/+lkjhBIVYstBwG9mYxYzWDLIL5wAjzKVnZNmxYfWj5+QHvj1zgaRQJkERuChZdsvn+s9 1LBRPn3iF0At6F+yfuQNVAuM8kTaH/kVUAwMo9c5DYGh7SZgEHzzu1wB71ohVPebWsF8LbWEWmEx AA409EIq2gagNSsGpoT3DcvRmQBaqnjzMg4CI4CGB22rJT0IreDwbo2fBM8ZTjKYn5rjJBBaoUJg jQ7+IHjBcAKshSVbJ4HQCuDwJe8mwa9MM5VSAOxkFvy2dcXEGoZrg+CRaaYK/r0PUyzP/h/vIeGd tAkccSEdsz3ZEUZxYTVrt1LINH9kmql50K8x6XxoYASS13vod6kyZtTlNO8EuCPpms6ofVG3CbnP ggAXNRnkL6tFnb0OOWfB+/YXQ3A+9phzTsLw0gp71l/jk3NpP3ZyGiPVTA4943gxBu9yj7U4I4/r 6RDobBRQT6IHWBchrvBi7gR5D2v1McDapGP4wK+4BzxCeLFuXIAoJWZ1kKbv7WbC6p9EDzX3Pe5B 8EaWaYD4P+iIA1n5HnPqPeAxpDgRJ/gQQVh09b8VpRZgdWei5L75O2+GdxH95l3/XRvYqJkR83Vu HXMzIgfApXw4l3daIeXDuRwEinNpDgJhRoZzOQhU5xIIzKpeyeFcjh7+JJxLXE95EJhn/8KthIMY 4/6JZcmJ4vEoVb6RWZwVC1546+n2QcChWGAMlYj9vjdMCJrvJ241Pu12iMzyNdMreL6nUDYI7zTj fQQ3Cb5wS9YgFAtFH0OzadfPlh5tnmqlJTDH7QINDU9X17juoauV4RkOgu+XRsTrn+jVd4IzZ0af jch+gVKbpYO/evZ0AO5Jmpm9cIG2Kvp5/TLxJsboMK5uN32M3vp+wHLOktfggSFcwA1syJ7YHARv uR3zIIrYLrjgEQ1VfmKW4fFd4ah4c7iDHt3qE2YU8gkb0URS2AGdjrwW9l7Kx7WNriohiRj3O798 1cxiPiX2Az6DQFTTwa3X4BgnzzYEDwroGs6hMx78hbBSCkMWZv2J+8Fl/RPolUzACcRmG7YLNd63 PWq5wBveGS75YgR4pSAkpsbNqWqQx6pVB4EoO4XbHgstbmbZILtR3fEvoq6P9T2redbiaEvN3x0T /kamyMsKHr9vpmgxADVT5I8b9EqmyJ9u0C8yRZ7erlxlivyp5sPCmA+Cf+KZopi30Ook+FbEhGkL WPviNAaiVXLD6mqTJdIUD2PuT7doSaYIdEYEh3DxiZEpCpulfOBapUDg2tLEhbTFrm7Ok0+IPA9u ddZq1DEc1nwQvBfGGjdL90GapTHGs2YhXfRQIMywhyzMwhjbPQjQAOPwPFsqlNU8VYQp6Kaz2tlc Tj08iVRRDX0j0p/qQpzMeS+nJgZ5KBbw/rtiWeHhCAp9Xk3TDHsee6SiQc7hBlwgsH/LDZlH7y1d 8MEDIEJQP2EOg65BzuGRuJwvPrEb9Am57zWDror7XqJjIE5aY7zjU9R1s5fmC3kiTlrjmE9weCuL 9yUGKMNjQnvq4ePanrNPmJM9b/1wtsrIhFmGxDBLEYdnglu+wGyqrW/xqYjDjRd7oaJcxnPJa0Xc AZdBlnahiIm9didMf8sBV0BWxPb9DbGz+Rvq5HVzC07ZYe3Mb5vb37L35q8aAKKLDqCbW6OnYO+0 egp2EHzh9nikYAeBmoIdBNzpOlKwg0DEziMFOwhecXs8UrCD4IWwpvcUrBykOadgBwE3MwUWZG1k DK95frTXFppiF/sJx77LIGBqw9SA1T/qRQ8jR6sJq+KWc/HmIBAJVDxisebDnQBv+Nz5sKzbeeRo tUHixky/SL8TmOXGTAxkFk8iiQtKIRJGfSPsMZ5lDfosHB6pqY4N0pyr5HbFonESAvTYT6sMggdh j63dKKp/lhY9b5RRn4U9hjFEwodlCd3s1qw2o4ZuvZqFh8iTIupHYY/vxVxUTno8WhR0YYFFb/2Q +yCQKVawVI7MQtrjOxiMkuNtW6pMe5wI8EaYm3h7vrLnK+1BzLXtx/Cn9uDWuOyHwFQuYQo3MCyc jLHtl+W1NeP6BUPygU/CFkffnRpVDHjZNXtdwUF0HjZ/NYTa1C/sthg9s8UYiTGvPQehCNpAeB43 qmNlTWAI3egY3sqiv3glzRDNwMNz9N38haSKaxNswrMrca+tpi6oAkvS5gm2LzK8D9eqqeJxYzLH XznYKrpdBK7C9cNDjaQDnqN1eKo6kSHwrT7X8PJhk1g40NZS2K64gCqc6r4fGdi8df6SC97i1o5f KybTCbAabmWzPPVQ0gkrPNTwoONPiuOzKG8f9lOTg+AXQZDBYjujz8JhDfWmi8p7ux+KHIATs8BD 1/bCHoNnZbdcJuB+EAS9gqWuXvuhyOLpuj0Dzge8WXehGjwey86BCctQAjASVceLD1grPMtZDsDB JEvfK9TUm49+D1bkLO+Iw5NqWXHPdgI8vBPXDl5HXPL7fTBVFAl8RCpt7r/5hId3LviY8OzOhQb1 qeHpdaOzIff9lZuq/zzeEo+6NQVrfl5WArE5J+IQjMJpX32vEcMVrLqrnBgxPFy50+rhyiDg20dH uDII1HBlEHxkQMHCNg1dokHApYS12Vyd7cwjMsdO3yDgSzY3rAtGhvAotgJBgWfw87VJlAD6l35C HCPD7CHetwUCswIr1uTKaUpFgLWgy3N8QWbz0RCmXHRG9+ximu1vmV7Cw/29it4g+CySi7hFxsFg 6DE0PPdJJPEgoh1YsAR6IgeC94+sJWMQ0U53afwKLWYeQytVR6xDW5ni+hNmBDOUD7IHPChNxiC8 fIs32Ksuin5BicpKnIrBC0p4JFAdA7iH8+cykql7Mmqg7Y0MVPZa0Rri92r+dcJtkXpsG1228kER vHtNZM0PeuGOzEZF+UoGKq1vLw2CF+LNEFjZ/oLNeNqEYoGrt37a5FIOEWvpxdWqNCOzGE9o+kXu FPZNNhVtmFikGvZBBAp4PUFVXsal7Lp3qSmOXjw4EAKRkjNfbQfMysu8v+8O2GIA6g6YPa41KDtg 1rbrHTB7ur2x2AEbBP/OLVB3tybDjoTakXHD8iSN9PCdsGHgR1g+htNZaXz+ifTwXJioBOHk5NlJ q/QdsIIqIU8CYcPQneqK0S79KdOPqxCxyHOVCIRM+PC4SslhjCCneScodr9Fq34CS1fu7csQoYGZ zHQIt8UOWa+Sb0/XcGnKDi/aJAoouYWWvaeAEmelYy/koo0B7EePSUe72Fe3eIGffOHDagctNykq Q3fQatNB3x+s6spViGrk23zr6Qd1DONY5eoTZuyg+aLiCU/E2I2uqz+IfBsm5Gb7P0sjhc8N6Hjr z1rFNPH2WqbbSj+XOQjeyg203I84DMA9yYzbfkLBWiXREyBGyGUCbmHnbN+K1MeAKV7C6c8i4+b3 68yqrPqpSgKoB464WPdLtKqskt1r/XPtMXNuLvftUG3l9tdjYuOD/DpHOc73z2pyv/EWy228xQKk y1o1wWONljy6ku1YgqDWe7tZ3D7bq6qN379UCruN9ue8mkSvqVaO9lei/jIsJdDNo/0FuxaGNUPw 1cTRLgpB4B5B9Ef7d7ySBJ4VsvVo51e68H0zC9HFaP/Aq0Dg+2Zlzv9JFFZGfTXHxyv7Z6zgQNrf La6UZTfH970ouwxOP8h2tH8Wr6+4zVd9fAXzRjYf+HgnCkjgTpDOn9rx4w58fFgUVG5tfv+TqNO2 b8lN+RlWPALLkczf88uNrb9IEs/8pYUf8OoiaX8jKkfUrRH+8HL++KZBTurwHRY8TROen/kFeqx3 2oIqPrBn/bjGaP9Pfn3e7W+VaMOHiAyvhx7tD/JBFrzbaTTxd1sWq7p8wJT1nfUBD1kuolcWV+GB pd1U2TpMWpIfL6o092cxRvtPvMYCKi/Sv6zBjCU7vCpbh7e4CXMfZRmJvm+mChfTkcGrunOEacr4 8DynwwIOKrhQuRHh3mTBtwjKyajMxxoU5PdvZXXnhsdLhnBl/xCo+jn+b2UBCrsVolr56y8lYbg/ 2v8oKj/g9kA6T8+w2hJtqt4ifp89HV4VT7/s75Go7Kvu9HtRybyi+MJS9d8rMp/m9wMXb9+B0T/f /OnngrsNV0cw3HKx11Q07vbqa04XrkdXm4CziEIN1s3hmWdNllnuwcD4/SZLRTg6PVEvzJ2//ydR 5gGCjebPysGwZ1Sibhg8FpW0QSgXQ8uyEcfkjazK5rC83Wh/z+tUhP00oP57PDdfNfQY32/de1X5 9XpqUTecHj2jUI0GLx/353pW478Xf4DVdcBDsA8do9g0cOMmBVkc49rA19kgODu9eLmQOb0fuWzR aQlhkAqnaDi9o52XvRlO72iXZdj2esej/R1/ABGdNnCKR7ssuABOWy1H+5tFDTZwas1oF8+aZHRq 5/xEkbe7UwztZqX4RyFjrf9ef+3oX5Ye6Q+Dkfl/EE4zPj2WSP/ntZ/d+fcvRSGG2o+CjXZeKTCj 00b4+0o4xftzJ6P9BcNHwSsvKajzL/iCHDgeK/mb+zsjIQdVfhWXFumfVzGt/e22ouILneIaNXj1 XFHI3mjib/1ht6SKp/V33eoBD172BWsZ0d/fROFj/L474PFeVi12eJFl/F7WWsP5OTI/XhYZPzDZ J/wq53qhyBN/jHSKtfmDUwz4P5p/FC8MWqyook6/+7xFG51xvqNXld7+gGFShd/rp+VmVspp90rB MHmnDw/fVapTuh9XXjH5vfR6O/oP6cr+8Smgqiqf7hVHR5SrkV5xm+yVdZRLL4d6WlynympYIz3r 4++lZKRym0+W7A/djfaFV9wN88p4mP3Jk16ET0VXLyMzlYcowYdV2Vy9qfLFJ82LDh+8M+L1xe9K r2KjDw+fM6xNXfz9NcOk2wbXH1nyBzzkOyX4on3SxVuTo/ARrxeh8stFZw9qP6KbRdSwP8GkK5+2 n7fk8DHT7S4bsV2y0lvrhf1W/DG7W46nphTp728cxqbqNnDL8Tal0T7fXzjUZ+cdPsAypSsKl2FR gXDxeYfl+qZ0v6zKt3ndNHjvHF08/M0i7+Pp9x9lseS6JbY4Tl4/cF/Xff1dw7JaPPdaYfhqk2/q 4vA9p+018Jle1s3p4PD4qFOJev+x1/AzmmPYn0OJevd9V3ZKVzrlsSckVPakNH6/XLs+IfedLh0s /151v64/hEJ+/734fT79/s1XLCDAggYrDhVxp308g3InlY+W4y7h7EpU9LsHDWa0i6Cj9++O3z+J xwwx0xywff2WocVMdVN/H/vz0KP/ReVnfJo5x+P3ov9Cfy8Vc6/arLMn4fST7H4m2k8/f7kq2Dyb n/M0ObpcO3OWHn/GbYI8hfNC1GLGSijOaJPHqgoerZLNSiVmrMVSDuE8LCKGUIsqHHzwpB3Ck/XX 8f2wSJj/X7wdv0+Ex13miqkwtHrK93tEQeb3hafJMZcU5/e/WbyWgsfhNOm2uJf6Gu0/ipDi/PtH kSdHn3t+X3g1Fl362f9n7lTZPZWsLU7nzv3LNw17yGjU77t0+v2jLLyctn1xr1OpDtMlc3xvZUzR H3Uc+Pgsqy6XjchHvrFSe+no0S5ebgv7o44aPlzYd1onvniqfU8lq/zBvd4ix2dmweXYYyoFX/0y ZVmKZ8+0nz//WsYMvWq1Ih7TH0N3s10+fLIfrlPh0fd4vVHZj4aT/F6wJ2Olu3rA42mVSc8X8Mx4 xXrC46VoB/jHoqnPex3nRtSn4Y+i2O62KerT4THP1s7jo/DAF87z5O938vnDhnWkOH/MdPo9lnBS +YtOfVnbPrO/hxK3cLE8W+pPEJ3gfUrVw/Il8BSOiwXjS/D3WXr1bpvaQ5ZuznYjroV8GP2AzzJk 78+cEO0tHyWPaWoH+TZaf9Aw6PD2wH9qfd8Jrx1sP7Ht4vv4IGutKjw8pjzI75+k19/PfY52USQ4 YNV6dfWCVy/ReU7Vx404Bw+rYs0+aqvX9FrN6YL9MfU7A+r0Iq7OYNTxo3Yi7P9v+e6Jv4vPrB8p h+9z9p/ac5nsNwu33p5+//a3cvnmbz8ISbgKvnQGk7Q8BylfJ7yTnvIJhrjlo128iNIPaJSjnedi MRuGV11Huzjgkvdc/WgX76VY/L0ZzR8XD4znFI6fC6894eNpffjL8xe4RQzO6/F7jl0srOyP4cnC 9Km/75GP32+/m2zBZHLZij1KDHlgcHdS8d5FfzCn1aP9jXI4abS/5rLFPG+Z7Z+U9+O1/iNixzkz 2kXI1N97LMfvH1ZlsXtslJZ6HTOZ7fj1B+Vd+Dl6FpBl8NnJ1z+LXZiCNo8z17BNmNEuTi7h/i5o jdXkzb0Ytp8//1ZEXGnLdcr2u9XbNmny9o8iosIHeupN430vcNziwdsi9mDw95M7VbwgieEw6Z4F VOX887eLPZhI2j/ygAprpMzuH8Uz8CD57BazM8eDNpl0Lx+bQehkDbkQD3XpqcJ1tj/ro60c3KLx 4NAd0pGvQuLueF2u3COeKu6QjtjkQI8lz/FRtfT/ARwyNR0KZW5kc3RyZWFtCmVuZG9iagoxMCAw IG9iago4NTAzCmVuZG9iagoxMiAwIG9iago8PCAvVHlwZSAvUGFnZSAvUGFyZW50IDUgMCBSIC9S ZXNvdXJjZXMgMTMgMCBSIC9Db250ZW50cyAxMSAwIFIgL01lZGlhQm94ClsgMCAwIDc5MiA2MTIg XSA+PgplbmRvYmoKMTMgMCBvYmoKPDwgL1Byb2NTZXQgWyAvUERGIC9UZXh0IF0gL0NvbG9yU3Bh Y2UgPDwgL0NzMSAzIDAgUiA+PiAvRm9udCA8PCAvRjIuMAoxNCAwIFIgL0YxLjAgNyAwIFIgPj4g Pj4KZW5kb2JqCjE2IDAgb2JqCjw8IC9MZW5ndGggMTUgMCBSIC9GaWx0ZXIgL0ZsYXRlRGVjb2Rl ID4+CnN0cmVhbQp42sVZXXNTNxB916/QozPTrPX9wVuhMG1nYIbGHV7yYoIhLnYcbCfh53d17SvJ d68NDjhMQhxmj1ZHq6PdlfKFv+VfuHb4DUIIyb0S3BrBlxP+jt/w4YuV5FcrLpuv1VWCWu8a2PkG xz5+pw/RfKGPL1yAMpv/p1+0YyKNFDbyqzl/PuI2NFb8kEFwhf9Gcz58JUEgi9FHPghnfPRfBVNS gTGKbaED3rWrAM7E1tXgddfufWKCZpbM445ZiwguqDx82bWrCFL4rZ0Nrrp2G0B5k8dfd+1RgJeO F/psx26Eg6jK/Koz3ugI2hX/ItlZZfcSgnC79mq8FQaXH7LdNvaXI/b2pHsrjO3bXNdhh5H1uLts 3+qlxO0Jhf0NsRsFWufd/UDs3oD3fJ94pMTtiWV3/+ralfAgy/wvid1g9Ir4Rp3dlSpI8GV3X3XH a+lB211+rLZrDbbs7jkZbxXEQNbHst2hfPcfDmlwfl/GTys7a+wa5Wv2x984XcenY2fSRAtOZ/uk O97i6RVlf8bE7gRYmfVxS+wR11/it+janbKgY9bHjNh1qOM/JXYrQcusjxWxBwHV6fqN2KMDF/ee fukxu4UD++OdrP2/6eojCA1O7p8/SDw/ar8+Qjr9xf4vGe8DyBKfi64+Iuqnyj6/b7ML/87swg5k F743uygf/E52YZvsoo+cvZvb2IHZMVINOH3GVLmENd3s1qye1UjMXwpUyOhtgtqBqCZDF8i80cgO RFssEq5AptSLjr4RWgu5oV5MxFJnD3rBGIGu6K6oF6cViIruekndeGNA6BbD+mbyARNztaR7Cgne gdl4YXv4JvnJysv4kALYcdWNKEB6FL1oPlPvIr31XQVcKmO/JUL2gxR4Q4GpqHspvNlEqSC5Th2Q VRm9zZLJSQvBBguwVhTImngxWEkt9lkZMiFeLPYqRhda23RTe7GpXTG+QN4RLw5LZsScRifKXjxW 3YAdRobMZpvKWYOCxqDEaqrLAXEUsDzGms4fG8FXkGjxePbELgOkwBwrQoHcEB9SBA01Fbok7GQC hGpJl2eNm8dnsywkdlBIGy07PK6HtMx+uFnMFBil0Gi5h8JrqmXs1/DksxY9mN5QpWJNjsFlj4M7 KkNjIfjiZd0jVOzrgylO6JY5g72Pq7xQuTtvwQZZvIz2abl4GVOVytRExuLlM4V4DK+T7ADdqA1q 3RYvy32Ziv3A7haBsR2BGUwdvzhZ9lD4mwosNXy6oMt2lGQpHai0rhbyvkeCEmQPgNUCjEleLaRH XlJAcKFAljRVmoiXywrSo8CQMrsqkBHx4q3GHtn1LLlSYDoNBTH7TLxE7HRsPVGPAh1e9rCb7Kzo NJdRKTbg9Km1SFct0rBdaiUqBbKfneIqCgYRfRT+XDycrxfPHnEUsSoJcN8+C6yPicJtaJjozOR6 vb59Nhw+PDzA1/ntLSyWn4bTyfrj+dX1eA3X6/msFCTHHzZUXlxUVC5efOtY4twMG17L5xxvzokH /j7j1/zimCrHvjcDdCUgMBV37gwdCRylQnYchSbwNYUm8GxwMVneT5bP+ONvL+wYGt7XNHSmsW2v K2TAiiGxWLToAaoBTtlhi83mpk8pemnywVdCU8rE0xaaqN5T9k711OlE97GEwrJdkMYGWsbCEs/X MTvOHhHLSvuxSbnsQPY75elL0o9V+d0evn8Wi/njdX/k8cNGq3Aosl9R2QsHWMi22MHtZHJ1vbp6 wpulj+oxvVJ6sf9JvVIPg/m+XrwFbx/ndjola7Er9RmhaZuNd6doZUZw2ilZ1TyAt4h3tMVxCoyP GUEvpl559FGY0oYNLRgZd4BHUAZcAVzQ7kYkGjYj1rRFwjwVQsiIJfXhNXh9aCmYjwM4lRHjnpuk 0FCFdE59KCHSYxFZbPGhNN5tMqDcnCsfeJOPZbWz6T31opvm89BqjPEgqonOnjAluuam/StToque b1jVhT1RRnRYjlz3BYkNhiQjOpPOj2zBg+H99MNkMRu/P+UNUgiWy6wwgTJtq+yJUmJZP8NmxPXN f9fXiyjAnraN1GI5+bRonijYE3Qjxuj9Ydoi2QaJp1dllpMPd8PJ/eRmvWo6/uaHUw7m+u4J3wis cL/4iaCHwfOe51S8l2O7uQUPZne07jkJToYMoVdhi9G1QmVEX90LEKtpaMFxGAifSW//8rxT9kSi YQ7Q8M6BiT1EWf2Kimk8I9Y9b6hYs2Ihuqou9f8DHn/DmQplbmRzdHJlYW0KZW5kb2JqCjE1IDAg b2JqCjE2NzYKZW5kb2JqCjE3IDAgb2JqCjw8IC9UeXBlIC9QYWdlIC9QYXJlbnQgNSAwIFIgL1Jl c291cmNlcyAxOCAwIFIgL0NvbnRlbnRzIDE2IDAgUiAvTWVkaWFCb3gKWyAwIDAgNzkyIDYxMiBd ID4+CmVuZG9iagoxOCAwIG9iago8PCAvUHJvY1NldCBbIC9QREYgL1RleHQgXSAvQ29sb3JTcGFj ZSA8PCAvQ3MxIDMgMCBSID4+IC9Gb250IDw8IC9GMy4wCjE5IDAgUiAvRjEuMCA3IDAgUiA+PiA+ PgplbmRvYmoKMjEgMCBvYmoKPDwgL0xlbmd0aCAyMCAwIFIgL0ZpbHRlciAvRmxhdGVEZWNvZGUg Pj4Kc3RyZWFtCnjaxVpbbx63EX3nr9hHFagYcnh/dNwECNAUuSjIS14EWYpcRLKlTw7af9/h7vKy O7OsHFUpHEMx5uzhzHA4nLOrh+n76WEyHv+TSik9BVCTs2p6vJ5+nu6nL96e9HR1mvT853SVoS74 GXa+4MTNMznU/Ac5HiYlwS7/zv9jvFD5SeXSdHU3fXkxuThb8YeOagL8e3E3ffG1lgq9uLiZzuJf pot/djDQIK0FsULPpr0dovQ2Faqzb/f2ELInaBbZfLkzG5Wkj1Aff9zbIUmtwmoXZ1d7u4sSgq3P 3+7tScmg/dTcFxu7VV4maOvD7nlrkjS+8atsF509aBmV39q7552yGH6sdjfbv7oQ37/q3irruM31 O+8wswF3VxxFrzVuT2ze3xO7BWlM3d13xB6sDGE6Kh6tcXtS291v9nZQQeq2/lfEbjF7rfgudrur IWoZ2u5+vX/e6CCN2/onersx0rXdPSfPO5ApkvhEtXss3+PDoS2uH9rz7zu7mO0Gy9ce59960+dn ZxfaJie9qfbr/fMOT69q+3NJ7F5Jp2t9fCT2hPG3/H3Y2z04aVKtj9+I3cQ+/++J3WlpdK2PE7FH JbvT9VdiT176dHj6dcDuFgf7E7zu+f+xr4+ojPT6eP2o8fzAcX3EfPqb/SfyfIhSt/z8uK+PhPXT dZ83a3eZntldxKC7TIfdBUIMm+4ilu5iP3P1fW8Tg9UxUzM4/wz5RCpn991tzo7okTl92AIqei3g HpJiXrGDXM810kO0iiChR+xJtI5W6uYVXlOEBIKT+LNCbimLwW4cUoOcKIvFhpJ7e1uIsDisWd25 O72hNMHYuW8XzK+UJpemGaYl4eGFLuh7QgLKYnY7X94RFshXSG5hBXL5jCoSz7whSRWZdWFjRJ5/ MA/kjvwFrHtmIYuXuTBZ6wTngnvFY9yvH7A1reuLfv3lGu6RuY9hzdSElb3uIcngqBBqQGdPBJLw NkvYzivkkULwlBgDjeUDgeBwk3JTrpCJgUQvnWkslxSi8WrzcRSRBjwlsWN5RyFGRZnS0BeTZ5SO 5Q2FWFwo+cbyK4U4nMR0l5drBhINzsLDiLzDeQ+GEQWjpetYmNRF5bAaxCjoiKc6dCxfUghWggyj LQIsWxm7tJwYSADpuoBuKQR7pvSdK+8ZSMBeJwaJA3Aq993dDjFHVHz+HM8cUdHlGjcWR6g/0KXE i1yojdIE3gUgWbIWxQ6SFfSrqp1tI+Nd/G7vooh4f0LqXPxEW5A1qIlSg5CyE1pple/YCrnlWpCd T2OBTJRFwzzdVsgT06VQ3/YZ/UBZwOAR0IOFsJFh1M41yN8pi8GoQ8fCHPssVqIf5sVi0A6GETlY +ssgLw6D9p0vb7lGhh1z4C0eZkhSdTH/xrS6vE4cQJAlYv6HuY1Z4tgGuaEk0UNWQcebmJshXlgd yyNlyZNqsiNfQOEOpS6gbwkL6F1afmB63S6it5TF4Cb6Li/fURaLR8R11fI7ZbH53VBXLfDZAkI8 v6PpsCyMP2OCLM3DXsCst2GHTM7jUGMruuxLD0n5MrSiQi4JRKs8a7Q1S41sIDg+5SmhQJ4oROO+ 5DuoQM5f8w7SQbSFvd3ma22v7xkfExLqghZlTOkh4JzU0yhQiGHuaIXk5oXCcqmT/14X2oBiAz1f lOUG6lJ+c1PDONEwrEbx3kX6cRE/Gwgqyzw+FgiTLmdDfgXXQQiL1zhs+Qa5oiw+6PyirEJuKUsw bm4eg4hCsvOF1RYiLNGB7MplLdItJGlpO1/uKEuKS2c+PnagAIvEjHwBFZf5sW0AYcG5Or9gq5Df KQvg5K06X4CyQDDz/N6CJiwmv2nqUqcoS5b/vsuLpywWx2rX+SIpiw12U1JPlMUpVAEN8S9K4rzb pGUheckJXFuNGCvSedzFMhuOu+LFs2R1QVAX5nGXceFw3BUFve7pH8iS+Fzdzrkozv5GRTkKG59P SXGRip+kPM6pboYIXmUlFFnJ6sZyxYy7Js0zW2H5xMyy2sxtqrCcGEh+n2tGvmgs8LmTHUekAaMO nS/MvGXw3MfQWBitbPG6VdBYmNnP4mCnuoiYhRyWAHQsNwwEuwd0LMxCHoepLi0/MoiUcuFUEub9 QMB1TGosV5xsV/N0WFiYXUzazjd8YXlkIOv8Ptii3OChi+iJGWYx/9Bt0b8ZCN4kyY62COfdON9q heUbBuLzC5pR0UHWWD6NShcgpVmpHR8AbPBmnmuO9wgbfJhfVgx8cXp+dzsoF5ywUFjG0QGA/GWm bwz3o94lXvxCYx7BSu9KkUxYPzCtC8VBhAJekiU2bQmjtMZVxAPTlXKUqSI+EQ6cu1O+6QqCKdWs mDCmgnikHDg65ca28ZQ2rdRiuaMc+UM21umAwxojdfP0nnLkex1atJz4tnPtFMRcXeKVrvWd4Ira 7Td9o7dEr7dW8LHcqohLwlHUVkEciq2KeKIcq9YqiPPXfCO/0U++hCaep7SKh0dCS7BBbnTWNlF/ gp7MMospB0YqrCqreHhiFNQisgriIyPDFo3VEiUOJNYglavCKghG96wCqyBuGfG06KtBLKu8aqtM B+qqleUrjuU7ycbt2CM9OPkbT3c8mVzmEwzt8D0QDsgHuNvTJ0b8pfzav8tlq1zxv9YGW7mon1e5 oOdfPKqnj5F7+Xi6VjGeEYSQf7uhcjBiL//+SUyjTBnUwKZSMFLPWpNf5W0p/pwPGx7C//m7BuOB PvyssYI7mfe6SjTwKfqZl3A6Yw8m8oRqx2tdI7jkJiWbO1zh+MggUswzTOE4ZyYlLCUDleMn7ovq rBILB7OKWURi4ZgOpNsoFIuhJFcpOGmHQ1+XjneHmmzgxiLJCsWbQ0VWKBi1FTzeCJXiifvQCn0k zHAa88efUDm4LwtYHAFG2cgfFmLjOPqu0DgYDaXc/AX7OJa5ZatWpNe8DtON447RWFkTmMrRt/3/ AFE92hEKZW5kc3RyZWFtCmVuZG9iagoyMCAwIG9iagoyMjE4CmVuZG9iagoyMiAwIG9iago8PCAv VHlwZSAvUGFnZSAvUGFyZW50IDUgMCBSIC9SZXNvdXJjZXMgMjMgMCBSIC9Db250ZW50cyAyMSAw IFIgL01lZGlhQm94ClsgMCAwIDc5MiA2MTIgXSA+PgplbmRvYmoKMjMgMCBvYmoKPDwgL1Byb2NT ZXQgWyAvUERGIC9UZXh0IF0gL0NvbG9yU3BhY2UgPDwgL0NzMSAzIDAgUiA+PiAvRm9udCA8PCAv RjEuMAo3IDAgUiA+PiA+PgplbmRvYmoKMjUgMCBvYmoKPDwgL0xlbmd0aCAyNCAwIFIgL0ZpbHRl ciAvRmxhdGVEZWNvZGUgPj4Kc3RyZWFtCnjatVbLTlsxEN37K7xMpTLM+O1li0Cq1FbQpuoapVCo eBQCfH/HN+Ha18ONYFEl0U10jo/H8zjOnT7Rd9oGfgMiko4GtXeo78/0T32j9w/WpFdrTcNrvSpU H8NA29vw1PkrNXB4scadRjBu87t8sUFhWYk+69W1/rjUPg0oPyihNvxZXuv9IwLkKJbnepHe6eWf hmbIgHNGbakL3eMmQXD5WWrxpcdjLJEwrAp82sEWM4RkxuX3PW4yEMYtrharHvcJTHTj+osezwiR gq7hqwnuMEA2dX/TrXc2gw1VHwuuGjwSJAxTvFnv0fHx04j7AT9cqpP/Wlt0/qXihi46zmzk6qq5 0xNxeVKN/kbgzoC1Y3V/CTw6iFHPNQ8RlyfX6n7qcYMRqO5/KHDH2avNt+yqSyYRxFrdo369pQjW T+NTLW4t+FrdPbHeG8hJnE+NeOD2nR8Ocrx/rOsvG1wNuOX2dfP5d8G2+elwRS57CHbEz/r1nqcX a31OBR4QPI398Vfgmc9f83fb48F4sHnsjyuB29Tm/1LgnsDS2B9rgSeEZrreCzwHCHl2+imyu6Ud 9YmBWv2vfX8ktBBofv9EPD9mvj9Smf6K/xDrYwKq+fne90fm/mnc58PWXfQr3UXtcBc96y4mpjhx F7VxF//G3XtvUzt250wN5PJ0nBGy2fXuNlRPtUwfeLwojuzFtyHElhIRgchWysHQIy0lWQOeXWak HAuV7NnnGsaTEGFLDuBCpRghwm6awJlK0cdSxhoqEz1yHqWMM2xaTTBrqeLJg2mCuZAqgVvLtME8 3EqdxNGgb0ifpVCOFhjectTWZFoKu7yHTG3EgkJcSr+p5GAlD/pAkhzybeyqzgtbeZ7YVGWuXmA4 B77J8NWutlZvu7JFW/Plidz35ZnJ6Szv7M3MD0S9IRIiu659Jm9OqaYMvtcxjYx7qUFmcLZnxpPU oDR4T55484Rhyu0VR4aWGpzM4n954l8ThnUWYt3lQmqUOzJPDzuVcHwJYZW4kRI+BDA0Mq6lRih/ U+tRHqVG5KvI1aT/lhqJLzufdh0l5QzR7jpLGZaUp5GqTe/9A3f4ge4KZW5kc3RyZWFtCmVuZG9i agoyNCAwIG9iago4MTQKZW5kb2JqCjI2IDAgb2JqCjw8IC9UeXBlIC9QYWdlIC9QYXJlbnQgNSAw IFIgL1Jlc291cmNlcyAyNyAwIFIgL0NvbnRlbnRzIDI1IDAgUiAvTWVkaWFCb3gKWyAwIDAgNzky IDYxMiBdID4+CmVuZG9iagoyNyAwIG9iago8PCAvUHJvY1NldCBbIC9QREYgL1RleHQgXSAvQ29s b3JTcGFjZSA8PCAvQ3MxIDMgMCBSID4+IC9Gb250IDw8IC9GMS4wCjcgMCBSID4+ID4+CmVuZG9i agoyOSAwIG9iago8PCAvTGVuZ3RoIDI4IDAgUiAvRmlsdGVyIC9GbGF0ZURlY29kZSA+PgpzdHJl YW0KeNq1Vk1vEzEQvftX+BgkOp3xt498tFIPgEpTca5CK4poIElB4t8z3qS2184uFRJKoiR6b57t mTfj3chLuZHa8RsQkaRXKK1Bub2Vn+Ranr7ZkVztJA2v3SpRrXcD7WTPE3fP1MDhxRobiaDM/n/6 oZ3AFIk2ytWDfL2UNgwof1FAqfizfJCn5wTIu1jeyUV4IZdfK5oiBcYocaAuZIurAM7EJ6nFuxb3 Pu2EYZHgmwbWGMEFlcO3La4iEPoDLharFrcBlDc5/kuLRwRPTpbtixFu0EFUZX3VxBsdQbuijwkX Fe4JAroxXsVbNHz8kHE74GdLcflfa4vGHiuua3bHmfVcXTF1eiIuTyi7X3e4UaB1ru7nDvcGvJdT 5iHi8sRS3YsWV+iByvpnHW44e8V8y6a6pAKBL9U9b+M1edB2vD9R41qDLdU96eKtghi684mMO7bv dHOQ4fV9ib+vcDHgmu1rpvNvnK7z0+CCTLTgdMZv23jL3YulPjcd7hAsZX/86PDI5y/5+97iTlnQ MfvjW4frUOf/vsMtgabsj12HB4Squ152eHTg4mT3k+fpFmbq4x3V+u9bfwTU4Gh6/UDcP2raHyF1 f8Gvu3gfgEp+rlp/RPZPNX1eHaaLfOZ0ETPTRU5OF+WDH00XUU+X56/ezjYxszpnaiCnb8UZJR1N O90+DtWpmYYM289l9qEBaoolxw6sKJvBIzXFEZuUfKH87FT4gHwNVZT7TiVZCbGibDuVoBTgMDby dluVqAwgVZSHToWHuktDY0aF53KAaivrXoTrC65Ky+OuVzFKg6o48qrXsdyBJi8ljlSAmzSA0UVm 1VM83xNURPoCUOBJhG4uuxQtD8u9iijTZkyJGkKVusffHUfxUOc7s8jIi11PSs8tsSrTEYpRxE9O YsZVynKCZ92r0oC1TjTr/EMPir9PAL4N9wvrwUFexvYJQyze7rdYM5Vnl8Qn8uGOGTG0SU8hB4ZY /OoZqZG5dHF0j40ZISQvPmnInmHTXZQlPvQExzVDyhLbIww+iqG5o3g+ivVZY12V4w9NEnZYCmVu ZHN0cmVhbQplbmRvYmoKMjggMCBvYmoKNzkyCmVuZG9iagozMCAwIG9iago8PCAvVHlwZSAvUGFn ZSAvUGFyZW50IDUgMCBSIC9SZXNvdXJjZXMgMzEgMCBSIC9Db250ZW50cyAyOSAwIFIgL01lZGlh Qm94ClsgMCAwIDc5MiA2MTIgXSA+PgplbmRvYmoKMzEgMCBvYmoKPDwgL1Byb2NTZXQgWyAvUERG IC9UZXh0IF0gL0NvbG9yU3BhY2UgPDwgL0NzMSAzIDAgUiA+PiAvRm9udCA8PCAvRjEuMAo3IDAg UiA+PiA+PgplbmRvYmoKMzMgMCBvYmoKPDwgL0xlbmd0aCAzMiAwIFIgL0ZpbHRlciAvRmxhdGVE ZWNvZGUgPj4Kc3RyZWFtCnjavVpLbxw3Er7zV/RRCzgUq/g+bhYJ4MMuNrCSXHJRZCmehSzZGtlA 8uv3Y890k9PFYRIHih+QDVZ/rBe/qmL3x+m76eNkA/5oYwxNkc3knZmebqcfp4fp8l97mm72E82/ 9zdF1Mcwi311kFN3fxDDzL+B8XEymt3h/+UfNihTnjQ+Tzfvp6+vJp/mVfygZCbG36v30+W3pA20 uLqbLtI/pqv/NWJMrJ1jdRS9mLbrnHRweYG6+Pd2PcaiCZZVWb7eLFuTdUi8Pv60XeesycTjurq4 2a77pDm69fl32/VsdKQwVfXVybozQWeu+/PmeWeztqHim7KumvVIOplwut48742D+Wld9/P6N1fq uxeNrXG+F9yw0Q6ejYiuOmc9EcKTqvYPYt2xtnaN7luxHp2OcTqXPEQIT67Rfb1dZxM11f2/EesO 3qvJd7WJLnEiHWt0v90+bylq60/1U+26tdrX6H4lnvescxL2qXU9IH3PHw5y2D/W53fNuprXLdLX nfe/C7b1z2ZdkcteB7uu326f9zi9psbnWqwHoz2t+fFBrGfYX/33uF0P7LXNa37ci3WbWv/vxLon bWnNj71YT0Y3p+uVWM9Bh3z29FMEu6VBfGKgFv8/2/xIxupA5/dPhPPD5/MjldNf178Xz8ekqfrn zTY/MvKnYZ9/Htll+oPsogbsMp1lF44pnrCLOrBL/JO7b7lNDXaHp2bh8tMXvjbebdlt9o5qJYMD +yZapY8HoBWJYBB2jcjNnCOtSLJR53JKFpFPAiWjTKTQiDwJFII5KDVVZCdQQNukg68iz79KGEqw qZGZXu8lkAWvx7wILQfnRMSBmpwfGUUexe3oGtX3HoVotW3s7mwUg9fGVZSpY3jyQTcoD1Iio9CH 0KC8ETIMOp+r1QLz+rIjZLGVsWoV+uFLs1b9/plB/ThsbO3cb1G0m5qsLn5i51/y4DQqcLYdFaaL /x4OTiNpUTcziHeRPvZlrYjjgNLPVWR3yNVWxHtUB1dFPgsUnGUdGpBrAeJB4TFWxY8tYCsSUAXR qFWRw7H5coeuMVUdh1I8bIyfFvuRByVuHPqTZbPGVP3lRm9VQUkVHCR6KhxKWSvpmdArWLVIH6v9 iUiCjqCxVWT3IGRCDDrFijIJiYimkVNVa4lYKxIi6maoKI9CJFnWvmJIRVJG75acGmyTkYCRGntu hQgZNrpR5Kkj4WemHKhCaERxGEZOISoU16Ds9lIGhWiIwWBbEPIq8kGK2NJRjfxKzsw94yry3BGx XiPbRm7zZu68Rk7xIH7buLazkc9z/1udInMSPV7SpZddZO47jilVPDeOueuIxHwSoo5nUmkXGpRO LqRkcUKGQcrktWty7l7axAZjJ6dR1jFGOE22wvwiRTCeaUOjMHEZwNDgDvTlMmPZBmXfEUEkw/BA s2WLSW5ELmwRyBQax3T0dYhSGPqXPXp2E0f5yx6BdE3+vpMiASRlh8zAIZZOSw2SiiPiGIQuncKj /gLr18KjTljfkWB9mdSuOD1VaVihOoWBc6wi0uneWxApVZHPAiWUsDhbRR67pSNzs9HuRsCUiYvQ uq4yMjCJS5VqdvogUFK5V3DNTvKAZc5okVtlHgTMoThUkedecQAhNI7ZSxD8KkPuwL0ExtWx2eha orBJ8xlcRB765cE3fnkrUUp5cDTUxUajU4MiY0Su3MqFUajJZVe6/kZdgYJDo/PJRpL7Md/yyUZK lgfW1Gz0/hz1N5F+ljBlyAsnEZDEbtCRjE3K6GvaM9BJmJwxKro2AkpQP1of9m1+y+qAwtla/SRR CEU8Nwdp97lD7DyTzVFG9VibE+oQj6LEFtnrDr5Th/Zb8roh3Z7qDq87C7qtIPePHV5vvdKpUx7V 2bqKsf+bBqt5tmM5WM2z3QuOItvZrqPCm3ZaUiez3VF6YdiTwQ06liRlOXOpOriVdnaR+CRHOwxu 2VSt6gyp1sHNgRlNs89Oznax3JqEKvIsUCIamELkvE5/W5CISh5sAzIJkASSDqlKPAqQhIOQbGPP nQDJcEqKsbPPipIxTGRuvP9BoJDBSTCx47iacCamuckU6lYU8lzufqXjKgqlPLcvAxTGbJ24mwmL CKZgdFtDFGfcXLwWkXuJMh//JkT7F3xTImZ48zszPKb4l57hOyp8L/swNAFgQbVId0jQh4Bxya+A y8ls+zBL4OtcUSTtl4tz67midKau6BHU5CvMR9mqUdY+5JEyaE4xXduKIhu+7Er9jtPAajIggRAr Sq9Vc4hBqii7TttCyOXUeKY362df3mmNlMEAoqmC9Cb5hCg17u20G9YjShXk506PRWWKrCC9NiyV dloN/E++vENr8qU3phe2CUODAmbR1OjSa+ZCxnitRurGwnxNunTcHzO6xlRROqN+Qn7bBmX3ttOG gaopj7KOcoon+fLQacMQoxxGWYe23WnbZt11b0pHj9XYdL/rAHG5hrIVqNNlldeZsVHn1UvebW8I LG+nUdW7PfTkdEJTkM82yN6h5S+H/tiSysCUSZNWhF/6Y2ZaNer0xpFKKferRMfdpWWgEFeRX+WM OV9QLuuvJLEFpzNVNQ4vl8UNJtc9ZGbkgEmAqrEPAgMjaixHJq8zn+S9TOUl9Xk9DteXeaDH3Clw xfgkMWbOC4PAHjivWvtOYsyUVzFu+5Rn3cgfhfGYR3p4sqUZyfUSQoCUe8k0dEhAZGI83eYUI6BO UMXoVZuY5hFtYE3yoNW6zWX/RpI3qp5iZJc0xdPYbTgKHjF+YMx8G+k2sZOXkc2RepQYheW4PZaS 4si1ubz7TYIw6L9JgN41IzrWbQKcYtjsTxKgM2k65Krb5tnLT3FlkIxn5sgXfJ+0mSM7GlyJht2C IJ3zannn9k5Okcg7HKRp8A6xfD6CYh7PTqI+RI0DO9X3g2JAPMyQK8ZzZ4QsKZNWjL2cD5HdhirG pZRIrEtVXSR+kONhGVTROsazA3EuN/qOVww5BOXyuVSoejx0Bj84NVeM285QxzjMUZ33OlG5pYoD n5YXUTrwivFzZ+hDCY3VH7uORKEMdd5YsuADwyOIw9g4CC25omgeYsTyCdEgxdAA+1J1zqcHzR2H XTGmjgRGFKoulbcHFBEWW11615EIprySGOwSUZZctaUTuARr/RAjl3tkGoU2I9djGPgU9YLR06rz 1qLrDeWudJF4LyUItoRBWJhhSqOGDD5bmJIGic7WoxSMDLEwNdWgyMsNdgisywOHzqXC20FgMbBb 3XBHx13ze1C7MeWLyo36M9+kFLL3UZC9/AjDzjS8CnfIzc00HFaJZ/k1SJy/9VsEZL3wHg21qZvc CojAQQdfN5H8GA36JbQQ59WIFq1dY4rMijjTY9X0RmAkjKloRVcJeRDLKePsBnqUalAuNc7rUapB rsY+CggUA7AODdwxFwOum1xKDCqsMzIFEK7wY429wCgHMXBjbLcYhNAEX2C48qkkNdZKJs+29ceT xPDl44eRS6n0E7mmx28SI8Da4Ef+iIV2aBBaKv1E8iM9Ypz76UFsE2Ib7Ca2B0r4P/Qe2k0KZW5k c3RyZWFtCmVuZG9iagozMiAwIG9iagoyNjU5CmVuZG9iagozNCAwIG9iago8PCAvVHlwZSAvUGFn ZSAvUGFyZW50IDM1IDAgUiAvUmVzb3VyY2VzIDM2IDAgUiAvQ29udGVudHMgMzMgMCBSIC9NZWRp YUJveApbIDAgMCA3OTIgNjEyIF0gPj4KZW5kb2JqCjM2IDAgb2JqCjw8IC9Qcm9jU2V0IFsgL1BE RiAvVGV4dCBdIC9Db2xvclNwYWNlIDw8IC9DczEgMyAwIFIgPj4gL0ZvbnQgPDwgL0YxLjAKNyAw IFIgPj4gPj4KZW5kb2JqCjM4IDAgb2JqCjw8IC9MZW5ndGggMzcgMCBSIC9GaWx0ZXIgL0ZsYXRl RGVjb2RlID4+CnN0cmVhbQp42r1cS29dtxHen19x0JUNJAw5fHeXBAnQRQu4VZFNNrJsxWr0sH0l o/73neG9fJzDIWW7kGAbssG5w+HM8JsXrz+sr9YPq3b4W0gp1epBrtbI9ePb9bf1dv3h54NaLw6r Sr8OF0RqvUtk3x/plssv5CHTL+TxYZUCzPHf9BftFkmflDauFzfrT2erDWkVf6ggV8A/ZzfrD78q IVGKs8v1RXi5nv2nIQMFwhhYTqQv1v06BOFMzKxe/H2/7j1JgssLLZ/vlrWMwgUoH/+4X4colPSn 9eXFxX7dBgHelM+/269HKbxyaxV/2awb6USEuj/sPm90FNpV/pLWl2bdKxGk2643n7fS4PFDWbdp /Zez5dWT2lYayxnX7aRDzXq07jI6vVJonlClv+3WDQiti3XfdOveCO/XkfMoheaJ1bp/26+D9ELV /X/p1g1qrzrf2c66CoISvlr31/3ntfJC2618S7uutbDVut93n7cgYujOt5R1h+47vhzK4P6+fv6q WV/Sukb3NWP9G6db/ezWF2WiFU6X9bf7z1u8vbLa57xbd1JYVfzjfbce8fxVf3f7dQdW6Fj847pb 16HV/1W3bpXQqvjHoVsPUjS367tuPTrh4vD2K4/oFib28U61/P+x948gtXBqvH9QeH9g7B+Bbn9d /3f3eR+Eqvr5194/IvpPgz4/ntBl/UJ0WSbosg7RBXzwG3RZ2tDx5bvvsW2Z7I6aSsT0UxPiSGv2 6PZbsk5LaRHdpfaF+hQeWhKHCBNjQ3KefKQl8QiCkc6dSe7X+55RdAh1FOmOVMvpsrUkSjp0x4bR +nNPA1Yml0Capdp8Q6I9om7lctZTmIBhJ1QmPz6TaZSTnrXN8eZ9uwDFM5fHBGgM1Qrwz6qk5UTq MfjQ5c96ZAwWEN+CqyTvM8lSSGI06YpnkruOC0hEId+Y/WPHBZT1IujWxXq7g8ZoFkPrqnuHx1RD i2gqyeeei/UYEypFwtTliSyj/HFf/En5JqYDbm+YvxwlPFIu9BMAQVeZQp1t1zDTEs/gXCV5W0mW E4nH0GUrxUXHxGDwC8kRTiR3mWQpJFEL3+5z03GxlMJYhqIwcQFBXGtW2BOJt1pIUlImue24BIT6 iFYpJG86LiEA+lmjuPOOSzQeFRcqyX3HJWJMlejRheTq7uhnrTElajcERt5KgskR5X6F5K9HYN2Q uEDpSyFZey4gpWj0/1PPBMMRuste3C0N5WFQaBbGjJhKOWFmNlJWKwLVwuSeIcF8DPONwuVjT+Ik CHvcKEHz1QVDE5yYq8WrcqCFN6PymDs5mJ4oKIfpXeXCiBIwg29d6l1PErFG8lC59L6ARjSUJ/U3 rSFBE5lYuVwzNKg8AY16/+hJADPOk0sd1dsfG7SyhKOFzYEhwWPrRjO9CYBKR6+nXIyzlJ4VLr3y wGKBG81UM5YijK5cHnoSZ9EEjX6v38wC/vJ1dSaL6ieUhZiCfofqvQ9oinDkSSdqPOnSATLiF4FT Jum92lgyr6oka88leGFjs9HrHrOxbDO+2ehtx8VRZUjXp2605+IMKt27SvLQcfEK/R4VV0h6J/GY zMUIM1mCRtW1G/XQHyWpbqaWiHCsvJ4pNzrUXHAzEympSXUztWCWa1JsyCTnDPBDTLGhaq4PH5FU 14rbIz+k6qeQMNgEUaXoPFYuhiknQqMXJpQZSG2EmSzGGgzylYSDdUvVblHMUsNzGx20AD8zEtbM VpwuyTLAdY9B07qpkbxOnYvChbGAj1603nBgogPqTsWpLBF1JxtvYEJiRPQ64cgyQFKpgoDpiTCh xUzBVy6fmPiBlZxswOPqnKGJWLKryobZCVB3ALMjoWcqaqxMFAMa7RibI90w8UNroUzlwgQHEwOV ixOHAaulMGYvyxMVg+3GZFNMq/ax4bxLJ0tsOFEz3lhiQybp72CJDZmkxwMLKC+hSibp8cBi6Rwp EckktyzwBw0zWRym4hj+KwmDB54yzliOxOFBwJRTh81OHQmmnD4ej7QM8t+I4GRjmJ1JScwoTahs OGDHjLJRDIMHSqFf6MqEwQPMS4VqjMTgASidbkbm8okhSS2Qxox9yq9T67gwYfbRntpD0wMZjXmg m2qFCgff6PY1Xzk4N1WLjS6VBZkLVxXYY1zNXM75ssC72T1SWCAKZyoXDtXNMTZMtBvlMTZM9BKN TKEsc+FyfizjVcPlkiFBM9qGy8pXDrHRyz0H/CDMDBiwtPCpuBubKMG+mosCmPrYhst7BvYx9WmB gVGLxtRH2+mBDLiNod9ykUFtDM2UZRbDt9UzdAEbQqpPCrowIdP5SFOKwub6OYsPDV2A6d1IY1Vo VKVmMluaFEL0leQjE4MihuNQSZjiwynMvFQluT5nqw+aKZ5oFmYni7YDZSubvtJ1lkZfMZEMPBa1 jRcDKpfvmcYThTJdufRwEDD1UoRemUsPTVicYCRrZOkPHc0pkhXFMKhCxYVq2AyKi0aW90yAgWMg y0wY4Ke+k2684YqLMGglOz1Smh4aP1MMhhgrGlk+8xGm9cxBhLF2KgpFGNNwueUjjIbK5Q0TYdB7 YSqKi2ngNXGX1HeKjesO+k6hcTq273QMDZnLAx9gpJ0aOvogYmPnez7AgN6b6JmyYxU68LpgsmO8 OrpSMyHDoEIheXbIxUSX12JSZSpBf3Es6gpCQ/KRbYoobWaSUOIrKe8aS+IxXACVYZnk6sBmtcaW My+MF0WgS9yc+cC0sl0qCTOTlU97ZSPuFdfJplcQMdGMYCk9hKhc7rm816QWXWbCNSt0JGzLTLik VpqEbRNJNLrrSdgRFyNVylgzl08c5MiUpU+4WMBCqdEKBzkhPR+ZqcVhXQdz3VLe69XM6VLeaxsu lxwsQWWx8nmxgcrinMuLY2tkhkmg9yaNkQ9j1MpcHvhu+QkdBvCIOa9O/ZDMhSHBMurkTaOEFpey N4062GBM6oaMDwRaHpsUmcsl3+qwjSwPXDf92KTIXK65jrsNqWs4NhEGbZ0qrbHHpaRXNlwGSa9p 7hCX81ottJ/ayCt0hsrkz+fMimU/aP39RZ8WY9YQaAAkzagdoml2TZCfSa74prwknJWnyRjTeFH0 hmKz08o0XgKVPdKM4rkj6wZbuazjtDhzYQDdGyciNc3kcAhKMUqrRphe3oBXQ512GhQMIWghTSPM LZMXe7zr1WIc7kh6pQSVyzUXx7wSqtHMHTeQPbY7xvpNQSqEmX7Tkzw900vqzaiGyX+5ICXFyRtG 40vtQDSezMYx8AmcJgei1JnwQI7n1JQ628bS53zqbBpZrn9/OUyepRkHslPyvDX10yMCxr7AQsIf TATDK6ZDK+PSB7ljL74gwi2XeMcEshkRDnzibWZ3MDV2gqpMVr6xA41lDvzQ2LpqmAd+aGwaWRgw B7zKPlQu756xTIj9fPUwrBKiHN2XXCTEdmyxcECeKT4xOE5VRLfHsi8i4nA0kmuIuJ8q1fcyWPqZ RgrRozN1H0MhWDsWnnqP0fQUm7ErsikUZ/2DG6NFxGQuDqeLUQYMn3aijujRa3ScyKGkdATccT8J XTbQLhmFbVse2k/kSNNUsDM5AFKtFOXoLVSapTanZXJsDalxP7ZLGrbqarnX/KwVKo8rRhCL8VBW khtujBrb4zIKcT61XeJ4zupl6sh3nlx5eJNeN8fJyDegZZydODuiqafu9cQyEWS7zWXPI1ogoOxs 185PPT1MGh8GpN0cl+uzUyu4nuW6twyiS5o/ZBIOR3WgF/tRjl6oYT0B9J2Ass3XAO3yDUGyAVp/ HMR1w0r2IYt/bFa5ePfYMxb/2KSy8ODzZRpU+sfmlBM58pjSPzqlLEzGQ0r/2Iyy8JiMKLenYcy+ fP0XZRizL20DKG+6fbA8GH9mAbGcesInzZt9o+IkvGw0tPyfXyUq+y/M/goMt/8dN7mljDwWO3Nh BrMxXR2f63HRC77Kg6seMBez1d+YhN0YehNaeDARwiJ222pNZo7glBTGFx7X7Dub9CI0k3zHPbOh B6EzhdD9UpO7g5Aa8LSFBVPYpWbpOtkjRIsFZGFxwT3BMZSF+MmslhqlehmjDcYHjP4zGACFGaiu PJh3MyoB40wOSMBYzcJNYelF8MQ9QGNqB2GiD6AWqatm4Sjo0dIMn4EapEEXHgeuz6SFqmdhKJxR osH4lRu+otbVBOOBEhkNO50+PbJSQHXAv/5hA6obTlXzbNYNp345oDoYzS4MZsAS1eDGM79cvGSS vvZ1hgxWBWEiKhYvthGk76hRbWIwJ3Qwq00CFpuZ4k/mRSjNh9tdupAbqQ9Wd/ncPwi11PD2Ezlo fBJM3YV7DmrTy0g3HjvSnXdqYlylME2R0GidfxA05UFfBG2Me+BHJ43Wb5gnnHjnN9bnn/rAxMmU lZaeq4z9Q1md8v3JWaxHBykE3EMgCPTyfOymudPkhk8Z0vtPTPwyxZvnSmZ8vcab5vTL7gs86cuR sHPPp68DMJpG2nTZSXj1DQ2f5Zv2j26joaXp+PRBO7jJBVYxpFdNbvhwAMs+oDn7mAeGdXNy6WXw TFXR1Ms1MMGMkRy9RXLDST1AsBTotoD2PMMSE+yXB6kT8TEELdwL1Q1F980EZ1Wh6BtwFrAIRyzM FHfsyyCNRh/LQRMQQB/e7LKNUZgOqrrJVf8g16M1IhTVcCWdOk7fM8XbjgcNP6DuwoQXRV9ON9tN lu1wJAhp6iaf+faZDhN1pPYZuK2g/WAEk9+NpEv3YMi6iVkUYKJv6lkuex7grGgMx9VGkf4zi0Lx 0PPQ9H3/yuPAt89MnJhFGYzqjeEYOSza1sfZWaylWcdEHQ6jvgwTT09T/Xr52HFJSP85Q6Z43fPw NnX/iyNztRFmbLKS3PRM0jtVmPjH8ZnqxE2Pj1TV5MaB9I7eERdJ+1c+aU7hYGJbQKClBlymEM/0 zV2AerZNzNagC0wvr/4HJcByLgplbmRzdHJlYW0KZW5kb2JqCjM3IDAgb2JqCjM3MDcKZW5kb2Jq CjM5IDAgb2JqCjw8IC9UeXBlIC9QYWdlIC9QYXJlbnQgMzUgMCBSIC9SZXNvdXJjZXMgNDAgMCBS IC9Db250ZW50cyAzOCAwIFIgL01lZGlhQm94ClsgMCAwIDc5MiA2MTIgXSA+PgplbmRvYmoKNDAg MCBvYmoKPDwgL1Byb2NTZXQgWyAvUERGIC9UZXh0IF0gL0NvbG9yU3BhY2UgPDwgL0NzMSAzIDAg UiA+PiAvRm9udCA8PCAvRjEuMAo3IDAgUiA+PiA+PgplbmRvYmoKNDIgMCBvYmoKPDwgL0xlbmd0 aCA0MSAwIFIgL0ZpbHRlciAvRmxhdGVEZWNvZGUgPj4Kc3RyZWFtCnjavVnLbhs3FN3zK7hUgYTm +7FsiwQI0BZJrbSbbqbyuFYjS46kpE2/vocjzZAaUlM7rg3ZGAv38PLy3CfHH+k7+pEqix/GORfU SU6N5nTb0l/pml58vxN0saOi++wWEWqc7WAvDzhyfU8dvPtAx0fKmdSH7/EPZQmPK7kJdHFLv5tT 4zspHsJzKvE7v6UXrwXjsGJ+TWf+Gzr/M4NJIZnWkhyhMzqWS8+sDr2q2Y9juXPREohJFDcjseKB WS+H5duxXAYmuDvKyWwxlhvPpNPD+puxPHDmhKXJfHIi19yyINP+crReq8CUTfp5lJNM7gTz3J7K s/WGaxzfD3LTyV/Nybsn9S3XpuZcO7IOzDp4l5w7vRBwj0/Wrwu5lkypwbtXhdxp5hw9FzxCwD0h effNWC65YyLt/6qQa7CXgm8+8q6QXjCXvPt6vF4Jx5Q5tY/kcqWYSd59Waw3kgVfnI8McovwPZ8c QmN/l9YvMznp5Arhq8/zr63K+RnJidDBMKsGeTteb5C9PPmnKeSWMyOG+Lgr5AHnT/xtxnIrDVNh iI9VIVc+539ZyI1gSgzxsSvknrMsu14U8mCZDWezXzhUNz/hH2dFrv+ncXx4rpgV5/f3Avkjz8eH j9mf5O+L9c4zkfi5HMdHQPxk1efbY3Wh96wuZKK60LPVRTrvTqoLOVSX8MDdx7WNTOwOpjpwfGpo k9zocXWbd97JkUYJpKcd0Mf2kEMsMjzoDLLtYuQEEjQLsUr1kLbQ4hBnIVLTQ5pCiw+eBZFB9rtC Deq2YEb2GDKj+00Jktp0SdMropclRgvPzMHkLvPeXFRAFtXbyaTolxLirElK6NsSEJDBUuT0lZAQ 0KOTmvJIUnDBpMq4qWCk0UzIpGZRQhSYcSGpqWjRGhU3O9OqAvG+S/pey24qrsnDenYR1+ieHIEf n3EiE04VXfs3OPy/Uos80gTamUBkUFUTDo03Q1KFzhpQmnv07O8eQnqI1ohB5M0A2RdatEd/0pmW ttBiTIBPdYJsCy0WQ5ZAGxog60KLRSFV2idIU2hx6OWGZ7asCi3OS4RgouhY0HMtseILmx16XmgJ nKNpZBvdFFpCnJts7dCDFqRM3KlCHU2QoJiSlUMnLbCV6UkfYQjTGAISZPeEMywmxsPGeCrsJwyq 6jgflORZPpBHmjDkAylN0EDUTDg02hyJyGFeGNKjeyJziMJdQHo6BbGYV3DbGiBNAbEqtgibtCxK iMfUEce/HvKhgDjkJqaHpGVXQHCtwUY6aaElBMU/KDdliw+4/7mpAwWLpDKJ5z57M4jgEnPOpCkC DmS49QyQ30sIOiduK0lLW4GAOlwHpzaSUmCmTlquKxC0Kp7ZsikhSkTqkpblqmKNViiPmZ6rEmJQ TMw0M0Z1Q/oUMyjTzGfWfCkh1lgm3CQzFo0zD7u/SohzGLgVnXKBV+jQWQqsVi9LUMDFJQuZTxUE ap8MZCKq0OD9yZFKeiWqPQbsqZCRGNJZyGzZlhApLXoYmUg21AUTp5uJ2iDjZTV39aYCQYPScirw ugEo8/TV1w7t5B6TxbiSYio9raSkQoX28Z2CGtAVzg3OILsEEZ6cqYHcgKyQtJQhYh0uj8eNSN1z jqMc+GT5bFn6xaGD2tiIezVl8nhcg0NuTKWUIhaVzYy5LQuliUeyU8SgUMYzJS1NBYJIi4Nbr6VS blFVmExKdk95sRs3fGl4teGTR5kwhCm5R5hWTLjMSSJZwx/Qx5cjRTcPIUEyh5Gsm3eXuhLShynu ltraBBmCnQwhiH7j8o1uyvgS8UVrttGu0BJfNnBR2yiFIFhXOtuIPtMsGGyo+qWgiwiOMctmfrmt xDdYz7m4K7UIsO5EgqwqPVyi8uO2PmC+FEEi4o3UmzHr94xj8sCLHD+A41PHuySoIF9xl3zES6Lc BGNMZ0J8kXHymuaHS9quF9svd/vlZk0313R/09LFZr3fblZ0cdOs1+3qxdhI8j8O+COeAsx/AE3k ERYkmkhOU8WCy5/nbys03bZXy2YgiTbrq6csjCOiYrmbIoo8ugRMuSoSVbGg+QRe1vvlook8XcSv m+3yn+aZeBG2K66FWcc54YAkB6SOLzt68Ozo1Wa1ou2qvcUZdnS5jl5+xsizobAcwQUzrreb2y7i dndt86Hd0v3m8LXdfsa3m83qarn+o7P2eYiO1sZ/kI6s/bxZLtq77RLsscySfwH0OSvMCmVuZHN0 cmVhbQplbmRvYmoKNDEgMCBvYmoKMTY2NAplbmRvYmoKNDMgMCBvYmoKPDwgL1R5cGUgL1BhZ2Ug L1BhcmVudCAzNSAwIFIgL1Jlc291cmNlcyA0NCAwIFIgL0NvbnRlbnRzIDQyIDAgUiAvTWVkaWFC b3gKWyAwIDAgNzkyIDYxMiBdID4+CmVuZG9iago0NCAwIG9iago8PCAvUHJvY1NldCBbIC9QREYg L1RleHQgXSAvQ29sb3JTcGFjZSA8PCAvQ3MxIDMgMCBSID4+IC9Gb250IDw8IC9GMS4wCjcgMCBS ID4+ID4+CmVuZG9iago0NiAwIG9iago8PCAvTGVuZ3RoIDQ1IDAgUiAvRmlsdGVyIC9GbGF0ZURl Y29kZSA+PgpzdHJlYW0KeNrFWUtv3DYQvvNX8LgFEprvxzENEiCHFk3ttJdelF25Vr0rbSTZhv99 h9pdkSYlxbFhB07gBN9w+HFenKG+4c/4GxYa/hBKKcOGU6wkxW2J/8Y1PnvfMbzuMBt+urUXVUYP Ym8PcujykTro8AM6vmFKuDz83/9DaET9SqocXu/wrxdY2QGFX8xSzOHvxQ6ffWSEAouLS7yyv+CL /yIxzjiRkqOj6AqnOLdES3dStfotxY3xTABGHi4SWFBHtOXj8jbFuSOMmiOOVusUV5ZwI8f1Vynu KDFM40AfPcAl1cTxsD9P1kvhiNBBP/U4inDDiKX6IR6tV1TC8e2IqwH/cIE+v6hvqVRTztUJO7Cs Ae+iudMzBu6xgX2d4ZITIUbvbjLcSGIMngsexsA9Lnj3U4pzaggL+3/IcAnWC8F3kXiXccuICd79 mK4XzBChHvJDMS4EUcG7b7P1ihNns/OhEdcQvvPJwSTsb8L6KsLRgAsIXzlvf6lFbJ8ER0w6RbQY 8TJdryB7afBPkeGaEsXG+NhnuIPzB/s1Ka65IsKN8bHNcGFj+1cZrhgRbIyPLsMtJVF2vclwp4l2 s9nPDFQ3u+Afo1ms//c0PiwVRLP5/S2D/OHz8WF99gf8S7beWMKCfc7T+HAQP1H1eXesLviR1QUt VBc8W124UfJBdUGH6sKy0m81VJekNs6wQz9W+1J2YEkwCR1+SwsWocAxqX4Xw/6DJD5IKmcgfYM0 XB8oEdHOEScjkTbTYiBJnK9iJ5Ey02K1zxMbRIpMi9OSOBaJ9F2mBuqxIYoHGdw3mSIoamxIqoMQ WuHzXEZaSVRE+dPZhJChdqieIDSk71+5iPXZObLJdXDqL0AWdJS5CIM7mJklC3OmoYsQQcttLsKh E/CpHJyQiQhwlFVBy8RGkjMSKeleMmThzEPIwm/fjDEDDklC9h8u1fdyGj2TAh4oIO7EJIXDnRxJ Qk+miGRB+njteCWjCNzrHFJiFOkzLZJz6MxYECkzLdJCtDsZRNpMi+K+xKglLspBOnAXRIpMi1Ya gjQ60fbQCcQyhkHGxDu9y9QYuA3ByA8P/UCJ5b4fkkt2scq3PCamixIRaFjh1o5Mt860OH8x0mij 60wLo3D5xly6R9we6KmRRg/C8BsJQX1nwLNgF5w+8gJDz6MAwSfRFIXzviy2Vf0v7q9KfNtU63Lf VnWPN0VffC26MqWHnt69j/RQTm+wkHTsRyyEnkEhWAhFFpqk8L7Z7dtmV3Ulbi4HM3Vle1u2uLvv +nL3Blf1enuz8TasjjfY4/uRWbLou+6Eaw/qd0L2urz3RHZFX7ZVsSUvSSfOLF/NueCL1Rw9e/7L IiiiMFTzCQrneTU3cCkyhU7SoQ6HUi1ginBuVDhRh6Xx1dEFLbd5qRbQa0MzvLCR78eM1mhhIw32 0jDQjyI4F4Ghi4qg5C4v1EoQHlOpJuowdFE6tstVLmLg8nF2iYuDFsoxvmQX53sbf7GcRJpMBEo1 Jyw6UpVXfMaYJsIumZfBPEJ4ZN59LuInaqaWXA1DdWSVqs4FhFEkIpLblkkwC4+s0k2IaE2UWrIt lCcwbnScbXWdC2kKI7hYNIuG4KURm4mtDPSoMtJSTIgYP6umWp5QbdAPNW6+2sg015HP9bOcoxXQ XMhFYzjK/UjqRdCM95yQfq4btUwEo4NWyJigJefCKbSRLAqT/YQItJE00pJbnTMl/QvEQiRx6Owg 24OWKRHwnY7qQV4yuFAQ1BGXTS4iOSUmC9mXnx6gekxX+8tqW77k+JA2ddSZn93UTVD4o73ZlNDE 9Q3u+qZ9JYMwaqctcsy5iDhjQAHG+pP0auirfM+Jy3rd3u/7cuPZXzbV1kOXPf56/4qtlYO5OjlE W+6a22J7tm72Q4sFht2WNdCFZqvrSvJK7HzUWbUcdC/QaCX2mWBwvi/L9RVoe6WJwUKnPkGjyGLN SZiplUBH4RWGbr5v1s0WXFjUm6LdvMFdg+umx+ttWbT47qro8V35imOHkVm41eUhAzYNLr42N5DL V1X3isOYhovq585iEwzOD4PXOkxk61cyif+uMcUoDzcm4ZbnFh2FVzWGoLqOR+uqvmxamNCqpiYv eF8e34eH52EwBVVPeWp7xgPI8ZHRPzorhSb2P0xmQUwzShTMKEfRo21HHPk3K8ftiO+S9YbDOABd 2gkvU9xwGOpOTI69SgQ7aHdk2L5LcN9JKuo8jsKnmwhX/stGWN+nuIG2HJqlA45WNwnuIG4UZ3PH x85/eNU60o8e4sCfMjfiKT9GoSuEtumED0MXinE1fPo84XW63r9aR+c72Q+NOHd+HJvjD32b8o/r U/b1NBj3nyYdnvMPDGoQH8G9+yQ8YErT/j18dn//aZOLWf/CEAeNNZqFJUybKpj/LsNhNjZyDI8m xRUXhAZ6bYYrCvE3hscmw50hkXszfv7LolWR+VCCaz4fG/6zouKZ70ff+tyyet43BsjTsP7P1LcW cpvPu8bnTuTa7Ox+RLPBNxl9xxkJy28yWBtyyDyUPvv+DyZ+UtgKZW5kc3RyZWFtCmVuZG9iago0 NSAwIG9iagoxOTA0CmVuZG9iago0NyAwIG9iago8PCAvVHlwZSAvUGFnZSAvUGFyZW50IDM1IDAg UiAvUmVzb3VyY2VzIDQ4IDAgUiAvQ29udGVudHMgNDYgMCBSIC9NZWRpYUJveApbIDAgMCA3OTIg NjEyIF0gPj4KZW5kb2JqCjQ4IDAgb2JqCjw8IC9Qcm9jU2V0IFsgL1BERiAvVGV4dCBdIC9Db2xv clNwYWNlIDw8IC9DczEgMyAwIFIgPj4gL0ZvbnQgPDwgL0YxLjAKNyAwIFIgPj4gPj4KZW5kb2Jq CjUwIDAgb2JqCjw8IC9MZW5ndGggNDkgMCBSIC9GaWx0ZXIgL0ZsYXRlRGVjb2RlID4+CnN0cmVh bQp42sVaW3PbNhZ+x6/A27ozLor75bHttDP7sJ3t1p198QtN0TYbiXREqh3/+z0AJQICSDVOIu84 GSU5Hw8/nPuB8hH/ij9ioeEXoZQybDjFSlK8b/B/cYe/+3FguB4wCz9D7aHK6AD7dsKhx0/UQcMP 6PiIKeFy+rv/g9CI+iepcrje4R/usLJBCh/MUszh990Of/czIxRY3D3iG/sNvvsjgXHGiZQcHaE3 OJdzS7R0J1U3/8rlxngmIEZeXGViQR3Rls+P73M5d4RRc5SjmzqXK0u4kfPzz7ncUWKYxpE+OpNL qonj8f08e14KR4SO+qmXo0RuGLFUn8uT5xWVcHw7y1WQ/3SHfr2qb6lUS87VGTuwrAHvorXTMwbu sZF9V8glJ0LM3t0UciOJMXgteBgD97jo3X/mck4NYfH9PxVyCdaLwXeXeZdxy4iJ3v05f14wQ4Q6 54dSuRBERe9+WzyvOHG2OB+a5RrCdz05mIT3m/h8m8hRkAsIX7luf6lFap9Mjph0imgxy5v8eQXZ S6N/qkKuKVFsjo+XQu7g/NF+fS7XXBHh5vjYFnJhU/u3hVwxItgcH0Mht5Qk2XVbyJ0m2q1mPzNQ 3ewF/xjNUv2/5PFhqSCarb/fMsgfvh4f1md/lP9ePG8sYdE+v+Xx4SB+kurz/bG64E+sLuhCdcGr 1YUbJc+qC5qqCytKv9VQXWZ27BI79Lbal7MDS4JJaPiUFjxCgWNW/X4I7w9IPCE1uI8naAhAlEOk JJQlkL7QYqSCIpZAdoUW66BL+kJ6gjSFFgbHIVpFyLgv1DAG8erzOfIt1DAloGNFSF1q4UCYJlrw j6UaCZUhqkELx4YXUSJ4VNOVEC04GDhAQgEYSojhIpSQC0cyWoL1opZNCbEaEl1etK/T4IOoZF8g OIVhI/VAVUKgY4SEP2kZS7qcc3E69ErEcAF1kcuopjQdlxwGp6hkuGbyMDMlD3z6sZAZZfLkuYeg +bvqgr6QAg4UEHdikcJ/JiNFJBZ+uFR8Rp/cDkpOEJhdCbThCHkptCgYMmiiZFsVWpQzUMcjrZvX t1Ra9Nm2wFpN1FFmC/zGSh86xue8H7rjmS+Otb4qrAhNgUiVmKgrIL4bCmHQDNmUEAvzcuILXCAc g0wXMippdwWGURjbYW664HQYbaGocFSGTgKB6U1FJfsSwamfX6KSYQGiwwgyQ/oSIqCoqMQs3QIE BhmZRHFpf5gFYVZK7DIuQJwgJjlRu0BGgQeUvkgG9j3f1y84CQYz2IsSwyzwNTC7pz5a4GsMjM/q 4pEsY0RdNow1YLtES11CnJB+Dp8hHxYgYDvGLrkaajVMi4l5yVqSorcve0WSUjpVJ0qxENSP2ryo 2YLTa9ZsOoE9FQmIJQq/9COu8DBWY1vjh7bfNeO+hdfc3zTkicD5x7arbvFjVTf4qQni19swsbyt ti1yRhc548BZwXidcX5su6dm/7Jvu3G4Jpdjtw8TK7xbUv43TRd94YXBTADlBJRSiwSmzE5wYR/h AZsMdBGg/cIDG+9J2bHyJgAoh1BATgB0rLsR4FcyxaKCKpcrC5Ul0oVcRWcAC3O04jwC8jdYKDxW mwgIhR0lAAOtgdkIyM1gYRwwiYI/j3J0lDtoP9IPHWsUnFVQb3hGAR+XmLANMEF8Nz0B2kwDtDhO aGKGxwSAAkDBjHlJgYFmkBi6PgcgWDZgebfrnoD+GG5XZsBYAKCXWH8NscaBwwImkmjpC4CcGvkp 3roCAI2RXnAVDOyOcBHjrS4AmhJ34ZSS+iEzAros3hhMPUS5CxSk0adDBMD9Ta7CXxPKSySUosTK lMR5yDI/oaZ5t/nEARF9RuFPXqtldM6lwsG0Nn6HWw8FbTXhMVTywgJ7IozgSUrdf1MgIOA1Q+t+ MFC90rTd5AALaxljMdqKcLSGhTHszA0pwF/I2PXiBkuo9Bc66xydNUSkleU8UmBHtf7GdNWOnErj r6zOQwmlABhkrFinwCGv/aVmURjQDIASrJNDPOcagARxi+GMjgB/r5eE87bQAKO3VcsAFAA23E+s Vlgu/EjHVl2FuIARlSYannINfgOX+RuSjEJX68QCgjBLqH3eiS0PozxA0dL5NXeEOj3rKju1ocRo fgKgolMbCGVGbdSQZ4PRYtr4ToCu6MRQvH0crHGwfmDW8cDTKdB5o7U+I0+APFgdh+KtEw4veSuG uiNowiE3pHOK+NK2cMxTK5bMx8FJXnQ5qmVYQZbshKZbu9BhTvKiA8FEBfvfqqGhEztI+sQK26IL gq+kyA19nViFqPNS+NB+7+CMvmXv+Aqrj1HoSMD4AWiBwG+Hhz+aGjaP4UOzwWOPh+oV1pDusHto 9rh/xH/1+82A+z3848vzvhoafMU5PxqMS51aTM6Ep5CYgQjGHkPCF1js+L1mPzTdFe+cIsdgUxir fE1IKT684vG5wX82+/axrWGf6zs8vA5js7vizaSnhQItpqQ58Trz9e0nWAV9wevnUEcwctPPWbG/ nlsmAuisLzTDYTviXTXWzxDr1VPVdsMYXBXc1bd1E1bZ88sI9FUXyXSr9pcRjop3rgnpdYjf6xcY /LsfhmYYfNxCBfCXEv226RIL4Xe6d2AOuuoCweNt6IREExJarBEn8E3fDLjrRwy52I5Ds318x9sd 44qLkqarHrYNbncvzX7ou1AT3vHKS2vxf77xWmDwfV0f9tWYZh5+rgY8tLt2W+1x3XeP7abp6sb3 peodr7r8ZLl+03V7vIsD3VV3i5uxJrf44TC+V07A6rXE8B4mmtmh6Iv/y8ylwsUc10sU6jQtj0i/ 0HN0BN9U3T8gJxt8GJrNFb+Bzx0qZfG1zKHbQonDr/0Bgxs9p7bbHOrjEPTSVB9w0/WHp+cQfNtt /9cVrxdzvkIW7bOaGldC4n/LKVNBCmVuZHN0cmVhbQplbmRvYmoKNDkgMCBvYmoKMjE4NQplbmRv YmoKNTEgMCBvYmoKPDwgL1R5cGUgL1BhZ2UgL1BhcmVudCAzNSAwIFIgL1Jlc291cmNlcyA1MiAw IFIgL0NvbnRlbnRzIDUwIDAgUiAvTWVkaWFCb3gKWyAwIDAgNzkyIDYxMiBdID4+CmVuZG9iago1 MiAwIG9iago8PCAvUHJvY1NldCBbIC9QREYgL1RleHQgXSAvQ29sb3JTcGFjZSA8PCAvQ3MxIDMg MCBSID4+IC9Gb250IDw8IC9GNC4wCjUzIDAgUiAvRjEuMCA3IDAgUiA+PiA+PgplbmRvYmoKNTUg MCBvYmoKPDwgL0xlbmd0aCA1NCAwIFIgL0ZpbHRlciAvRmxhdGVEZWNvZGUgPj4Kc3RyZWFtCnja vVnJcttGEL3PV8xRrpLGsy+5pBKXXeVDnDhWnBx8oUBQREwBEgHK5b9Pz4AAhhgAlqMySwsl9kPP Qy+vG9IDfo8fsNDwSSilDBtOsZIU73P8Ny7xy1c1w1mNWfioMw9VRgfYVYtDmyf6oOEDfDxgSrhs f/c/CI2ov5Iqh7M7/Os1VjZY4YVZijl8Xd/hl28YocDieoMv7At8/W8E44wTKTk6Qi/w2M4t0dJ1 ri5+G9uN8UzAjLx5NTIL6oi2vL98P7ZzRxg1Rzu6yMZ2ZQk3sr9+O7Y7SgzTeKCPTuySauL4cD4f XS+FI0IP/qm3o8huGLFUn9qj6xWVcPu2t6tgf32N3v/Q3FKpppKrR+wgsgayi+bunjFIjx3Yl4ld ciJEn911YjeSGIPniocxSI8bsvt2bOfUEDac/zqxS4jeUHzXo+wybhkxQ3bfjK8XzBChTvmh2C4E UUN2r5LrFSfOJveHeruG8p1vDibhfDNcX0R2FOwCylfOx19qEcdnZEdMOkW06O35+HoF3UuH/KwS u6ZEsb4+7hO7g/sf4leN7ZorIlxfH7vELmwc/yKxK0YE6+ujTuyWkqi7LhO700S72e5nBtTNLuTH aBb7fzeuD0sF0Wz+fMugf/h8fVjf/YP9r+R6Ywkb4vNhXB8O6idSn1+O6oKfqC5oQV3wrLpwo+SJ uqBWXVgi/VaDuoy0dYYd+j7tG7ODSEJIaHgFVYGUUiA5kr8/AoEAxUeo4Ixw26Nh/qAEYluN6yBF 6kVyEdLQQR5TL4qBEkZeVqkXzdpi7SBZ6sXQtl86yNcnJBz935Ay0x7MDPLLAjPKjCP6iUv1xJpD z6OAuRNoisLHNpARUhhKLHRNhz7KUoCgFiIF6CJUSQ8pssSNgt51ig2YvIOgHmIlsSI66T7xArsR LCDRSfvEi7YWhkxMpkzcGAN7CEyyHtMEKYohlsGss3qA1IkXC5FhzA6QVhBjiKOcKMEng9dBNCwN LDpok3pxXvdcfNDYC6OC+6z3kMfEC2wfMH4iuqvUC1OKSB2HN/HCnPRDpId8Tb1wmONanmQgcSOk IDLK9W3qRvr2FEvhZVKGid5DHlIvioah3UMOqRdl/FxeDIwWYfT2kF3RpH6gMolWi6ExsCK6xZKB KUdPjkp7iVltvfxFhBMvToRZ2kPSJgBdt0S6JS6cwsbkIi83iRdOYSkyLu7rxAuDvUebpfLlwNev nlGfJF64Mn67XeIiOCwgKm7rxIuACudmKddcckpiwZsoX64gSU5PqFkEMYrYqDbXwcvT14lZaUff lHZfr2ws7SfxClonIKLWUzyiJwoWlInwtgHYOHed+ksbGqxDrCa0H/Z+Fp1TptoPq5lwdomKhnCZ 6JwmlX7/fCCjc7ap8gsJK2J0TjsdTqWf+znEBkyqTdb6OSTjsCSiDbXI4pO+TIg2PC8LG/NNRRse ibmO+GbTqq14h0ETt8041LSczGIs2pYNTibOETATfWPMZ5oJAzXdJgnNyI7XbMcWuUgjgvJ3XvYT sg7PGiwKXT4BMWGd771k87LeeakmFJtSckzjHBcjoO6iylxPKDaFumODl4mb9k8vUix1ALMwe010 RxN0nXJh9nZeDhOKDfWtFqPLKTwJx4WJ53W/8/I4IepKQgssRZeD8J/EZaK8uYDHwrgFJrRWwEA8 tmNwQ77z0W2stWhpjaYt2L8KAa8SBsF4kxec/ki5jyhIv/VNUPhYFVl+vy/KBh/qfI031R4X67xs ik2RrZqiKl8+5vv+F1zm+brGTXVO2sK6ZErl+H5fNXnW5OvLM3GBLztJBkdlhJ7598WeAJoiIMx0 NNpSP4HCpsTg4e2Ivjg0uKwavNrtgExW7ddFeYvX+SNkv8ZfCnj7Jp/rBvSM2xjiiE5yym2yedSH +3x/VefZYX8uKqExoS0mG/Mb2oCeRaHLFArRmKDw/pDXvuN+wmciYu10LIqkuBwoLeWiR1/UuGgg I1BcX72KbA47UAh44+FQALc634OE1GfqUh72w4n7wNtq56s+LHJnUi4qkyp/7AXXyyh0Hb6rQpB8 2eNmuyrhW1XnPd3zCS08DCY9eZ/n2XaQjPoS5/BeVoRcF5sj18F+RrqWJ9Nsu3qEeObN6mq9alZe 7aovXuiaLUS4Ouyz/BjzbsTl6zMSNrCJjoXmIie3BGfA1PO831ZljsvD3U2+v8S76vYWRnJR+qby Y/nTi5/PRNcLo7J6URfRs/+ZtjTsfMAmGLytQzKzqqxBXvISMlr5KizqVnIei7pox1zIOZRqjXfF 5/yMaZYmSfOf1x/+uMQf3sK3f179/g6aqMmIJ5yX9bHtG0/4K1RC6ef0zTkJC5P80zLsgsDvbvU5 7u6WKHD0PdQc9iWgqs0GavecfLnmyZAqgWwkrXGf/AcgSO32CmVuZHN0cmVhbQplbmRvYmoKNTQg MCBvYmoKMTgwNQplbmRvYmoKNTYgMCBvYmoKPDwgL1R5cGUgL1BhZ2UgL1BhcmVudCAzNSAwIFIg L1Jlc291cmNlcyA1NyAwIFIgL0NvbnRlbnRzIDU1IDAgUiAvTWVkaWFCb3gKWyAwIDAgNzkyIDYx MiBdID4+CmVuZG9iago1NyAwIG9iago8PCAvUHJvY1NldCBbIC9QREYgL1RleHQgXSAvQ29sb3JT cGFjZSA8PCAvQ3MxIDMgMCBSID4+IC9Gb250IDw8IC9GMS4wCjcgMCBSID4+ID4+CmVuZG9iago1 OSAwIG9iago8PCAvTGVuZ3RoIDU4IDAgUiAvRmlsdGVyIC9GbGF0ZURlY29kZSA+PgpzdHJlYW0K eNrFXE2PHreRvvNX9FEB7A5Z/D4Fu0EC5JAAQbTIxZfxaCTNWhrJM7Ic/futYneT1SyyLUcrB44x MaqaXWR9PVWsfn9c/r78uNiA/1u11maJoBfv9PJ4t/xzeVh+/8cns9w+Lab883RLrD6Gwvbtxqde fuYauvyDa/y46BXc9t/0f2xQmp7UPi+3b5f/fr74VKj4xyS9AP77/O3y+z+bVaMUz18uz9Lvluf/ y9jAwOocqJ312dLTIa3B5WOpZ3/t6TGSJEhWRL7pyFbnNSSojz/2dMir0XGnq2e3Pd2nFaKrz7/u 6Vmv0YSlia9OdKfDmqG9H7rnnc2rDW19TXTF6NGsSYcznT3vtcPtp0r3hf6n5+rvX1W32vmRckMn HZ5sRO2q2e6NQfWkJv2DoDtYra3afSHo0a0xLjPjMQbVk5t2/9LTQcfVtPf/SdAdnl4zvueddg0k s8am3T/3z1sTV+vP8ilOt3b1Tbvfiuc9rDmJ/alKD2i+c+cwDt8f2/P3jK4K3aL5uvn5u2D5+XR0 ZVz2a7CVftc/79F7ddPPjaAHvXpT7eO9oGfcfzu/dz09gF9trvbxRtBt4ud/L+jerNZU+3gS9KRX 5l3fCHoOa8hT7zcRo1u60E8Mhq//t94+krZrMPP3J4P+A3P7SOT9jf4/4vmYVtPO5x+9fWS0HxZ9 /muPLstnRhd1EV2WaXSB6N0puqgtuhgR+lPA6FKls1fSqV8X+3rp8CTxSHT56wJmNo0ydtFvO53C uWycPsGKIaJyYwBTHUtIFMBPLP0qtE3QjOVerJIA04yFxvLhXiyTTVo9BcONR+3exFmMBl3i0bHM g2QxAKvf3sQy7okFwBWrPlZ5M2AJfgXTVln+KHmcQ4FdW2Ygr/cob1tlIG7ACOrYjp4kS3RuNbat cj9gQS2lcKUkk1BLiZ3LnWTJKZVofKzyKFhAO3Roe3W6YDBmxXypalSAXR07GHl0YJ1fHVypGpzD 7JXaKk+/0v17B1NXDmbi9mL8S9DRRB97B/sOnC+m/++LUH1cXYsA2Q5F2BDEzqnor8UU5EzjfvaS s5S/1kc0jdhYlsaidpZcpKscn8QixRsie8+7g0UdLN6Q1tl7fhKreAz4FsONFKWuEgxiAs9keRKr BIQdOrEXvRerREQ2ObIX3YlVYk5r4i+6EaskxB/Rpsbyg1gF0TVCjMsdZU/xiL3o/kEsQ5B2hdwt o84siGbjQNzGQXjVp8tFkNboAznAFkh8VvN5CasL6r1ahYBhYC/6dksZnMVpNKgrLSP6C2uyXIVi EW8LAK4st3IVj6atPTdKsUpABTGHOwLjiSVbip1XskSs1cLYyypLRldOfNNilYQR2F3aLaIit9rL RXIEwsUXhwsaoXcOzCbfiWWwNja/oGgwmHa4or+RqwCCRB2uzgXAFxxaWT4MVkmWoGxleS1XsRii uLhSR4h0ENCGK01jjEcfMpereCqr8uWOPCVSw51VrBJQjSFfyhK8XXmEuvst00/AyqRLP+9k+qHc QrFt5372vcwcGCFjSI3lo0g/DutY0xju3w0Sh1mzblK13NISB55FpKxwsAwSB9aLBhqHjNUR4uot e8/7Qd5ASGfYex5l3sCwBHzLckMJSzesTwcH1/IGVmcay8fK8mYQDDS6Dj//t4OQj7ao4epcSiOC v2mRMYUSQ2S7/u7ZYBlM76G+SY2Cl0VoyF60DjiiXzNb5NUgMzi9pstFXLQE8Osig2DsdWke1FUG wvqoSyxAFjVJucGWFsDIuE+5w7ZVBi8qucNcqigZrLFcW+XFgAXdKNmBYbLMgBXWbphqbJhYJngq wi48ABG5JnxfVxmwGONX7kcDzAOgA1U+dZlXAxb0ez3yWBb2NcJOe6UksCFRn6WuMpDFIWCE/nS/ fq1RIi1aYh9pn4QbW+oV2cY9QNcOTQR2094NoHdj5z0m4xOLWAXNFYv1xiIzoEdzTXgWlUUaQADM tMk3FpnTg6fuZm4s9x9ksMUk6Wx9kxrlhoSOjMn/6mQSOrLZ96TGDpYhoU2zPUlTywHDTvBtFXl4 OWOUNJ4fngzZARN/W2SAWanJwQxj5KUmUVF3JQouElau6kHEAASknm1oVJnYBCU3HMsM43G5BanL fPe737JUR8/4T5fqAxH+IbESGmEi29i5B7bhMJsEcomD5VZipewQFUBjGVTZ6J6Zv0giFJ+xliQk PJclBCjQ/uAYlOqZgknb/sCrIiJYm0fvaWDJpBXS1SIpoiEbGLA0rGSxhHNM2kGcMBr8GhnPw6DG xmrxekcIp9LKdzTASpisqXNVWb4dOSfaFnvRqwHiwkRszJWKkKT5dmSRjeEzpcslXMSIFbityCrb bOjkYBnAIJ8KrDs43rz5NCizMbGEODAohoMwZGkm7+tBmb3dqV0JkzDJJSbNx0GFjAdjrg8m48Ho eKVn0GarW+eyILAOqw3cW2WZ7bBaYOJ+GKCpuNWt83MBsKmE8vmOKlKaG0xFSheyOK0LnK2edjuo s5Oji8OrgykVcuBvknV2ub+49GkIeDLNYtQO7b4k7KvPB24AArh9PwBue4m8c4+aqwdwO1ikrVH/ NbRXbnXeOeo7tHrDWCRuo95qyOw9Ui8Bqy9PnYuDRRpJxOrLUWY4WJ6GvdWYxYa+fjZO1gzV8u6D SKQE+DyZziHj6wGao+wF/Ej7xIMF+Ja+DpYBENOYbLliZCZFxJdLp+dg+TDIO8HQZbQUV50QH7e0 mzHii8DVK1axiMvDQFoOCbeOgjAAdYKElkn7fpAzDJQUeHG4PsbSsLnYUHCM/oNcImK4sqO3sJRC t4EwUKFiONOVZHBxKHTZ7ZghfBysgtgtwaUs2ec1XoqS8xboD5b7l6MCPZS8X3luRVgqaYdb5c24 QOd6/jBOO46tMorRNOXUFP11Y/R+gUj35FQboO8NawM6ja8TjZoAnrQ0EGC7MK18ymMK88ZV3j22 t4XQxEs5eNDf93R0R4hBVYY3HUPUke5sK/2+p2PcsHqTgNWkjCHEArqmC9AoRKpbUHv8agyJBuEI Bx0rLD2DhTUmJuPWMWcMCMeMZaf0qlsh001oavTiOorRsRQBdop3/fNYutOIomBQO0O5o6NsMNOT KU0vN3qF2hkwhgcm40O/Aui0aiYkP0dVGKwpwHV2jgaBVYl3oxVoM4ZGFh076Nf9ChZsaSM2TfQM 1E1JavYKZKC8aS+EJMCz60o1SMQZfJkMqiu86xlo8rDtQu1VMGfwek2uM4cTQ6SBFC7k2eBM0Iam 06ZGbwLVcMxgXnQWZ0IomXRmcWU+K3tpDtXiImYDyMv8pCNCZOdHMu4Wl/R2zzddIYErVebB8Km3 uITZMV5ZXErpFJ5ue4vLVvNjEPaU49bannkFaLRIyGoqAmhXRrEqw/eCIcMRAdVIFYA2W7Lh9BWo 7YLzDoPrdwHkuoGt8CQY0CJPx6Q6OkKzfGEuYO1WnM0iHFZvoVy1zpQNBNzASFUcBgfUgPJhbvOY OzcgJHW1GRxuUtNgz3wXHnUVe69hBgceowe3SKGKQOVSmGpbQWldCZv+NfleffZoXpfvtcn94GCv BV+mHqHyiggcaICNymdkGFproE4i2fuxQn9E1GrJZEsbgxKQgWo465gMN4OMHug29WB47PNx6cyn xiBSvo0lvh70+85aE2JUo9kmetCRqEcZuwWYsWYyg8QW6DeZwzYGIfe4B0eNLpvZFkQ61h4TOoz3 WIKjRl06JuOTgASYIwyT8Q8nU1RfNCFfTVGNsG+eQ9+RK6gveH91BcVcYfD+f/Zmau0avD1YN0dQ 3M4zmnGu9JsRbkU/yF28U9XKMzpSrPTeTSLNybjx+qrYONpHrlsRaYVSo22v70FKNoT2gjroPUbJ PpVW/ln8rzNB3Kw+JzPSjUjrGhEW29yNQMO+XCXPzhZdo8DIPEiH6gDL7WgFNiKonBr9ViJlqpqm Z2tg6y/mGfwjGBzKSQxDbEHBTPVie9ZFylJ5hwPC8W0q00jT/VGmNcCOp0OfLtI4y/D4Nvhq6L6y 0vvoajzaTg5T2ywde1stQULPgEV7CPP902Rzs6RnP/fqLdg2n+U7IVcsZLzw3EZGnBG8yrMiha6f p7IVTOurbsTZ03Vwe/e/BB5F0U3b+vteNeS3vr3+jcSzkdomU9VgSKFm38y00bLjCoOT2xEWJaUI 06MDyklpGvQUGA1cPAFk6YLLRmEZjZ7dGpvnvRM41xlCodPnIWEBWh3vk8CoUO5d8qTIBpvTWmO2 GiBUE+dnQ5NDpqn+U4+gaRIttbP5w2cCR/VvRuSWLdMG9eQNMsuGBq0WhUsTuEPzFBmSOuh3oguE VVS09XmRTTFiBqTtouzNVEbH+gSSX2brE+CjCZu0Y87HUbZMpj4vWkAYMT1mnFTx3tnvska3crHS X4h0St16e5ZPnRs80eZKl2jPldvHZMZowuiQSZXJTEKaRsOO7XxeyoyHtfj8cboDcE29IuMZmoVq 6hVlNJhIrdg0weOYEcl+qnrvZeco09eGx/MiY1jtqcg+1CuOz25fhszUa2ykO3qmvj7jAdnX/P0u kH1N1Wvo+4PcnpcZE+NmcEI+xRtCTH4R1mlmwpqh/op6AwIa5p6iOxiSvzAfZcg/mfnIlEhfWc2P NyIe00HNooOJOazt7Y+TnDm1juSAbu+P5z/KLlDpOx7WIZs8tsDFqXXTzDpz/k99d0RjvannwaMk TUzaJ+2rPivaqXUBmj81VnvtKZ71fJ4GPyDvYtbd4yGw2hDeS5OaHwhPujALzgqop8qCp8xspuC9 g/7NMPMZNT0/+tYWmn5FB81TB99U+/hJ0NG72PuFfJ769y05yKYOohII0+gI1GyN3Ls7+6Cej3Nz +yDv81305fYRtWvy/UIt9v+a+SOIzN8bJ3WiraNfTxgXOwFtJ2Qq4ceHf/SLDvqsXbSLIt5/dIuO 51+K6yGLkOxYXyq3NJNcrM+/EbdDnnz3IPeJP9EsaWzi95mH4Lxu0t0N+kQ2+Onus80I51Ol965L uCHkxMTr8z5NsJllcvoKcYWhGaCD/nqIG8xUvoIbUqjalcCAPlmx9XmROE0uBcGhXZGY0DfoluF4 XtCtZfKpATDA0K/9VD3lM39ufH3ep69cvFBfo+P5N+n+JdI2nq5vp/teFrpxzWa4/Ja2raPxq+np hu17hKl2Qy71ykEXiS9uX7rOfNPEQF9Wqql1RZRfp7l2ky1TXYd2H2SxmylwHs8LWJQRdrm572Le K4lv+v6cS+JrsWGUuPPUurdq1531d0rcNFrYnn8UidtQcBTe2RI3wurU9PtRPI+w2sNQfyVxE6xO cWofQLCZeZ+4a1Bfq8EbzH+2wTt4/18mVx2Bl0SqS2y6LrUbx9fOugHpQdbbD6OsqW0Vrq83IxqG cUmFCWSJNMwd2uYfR1nT2EMUWU8j4opYsIXZ/Qh9NbltZTxzgfWWj/HzD1d9WXt5i4cpp5FlvBV3 M2gZCGmCmYRlChsuV7oImxQ2gqv01yLpJsI0lS6ToiVMM1UuVtt0AVaV+ySr7fLbLmFWjlEXzFfd itdTEyz5s2Occipi0ba26hNyqdVmhld+tyeE897UaT4jUy0VzCzjYsqy7XnRG3Y0RDhXjaep87F4 amtNW2owTvfufVhZ0BCFsseMzPxSZmwEfG15kZAD+p32aip+oB88mR8uFQrRzDUbfWl/hlmXKWZq f051m+g3UNrh7Z9ZngrxMgUQJoUuFuKZZrUO+k+yEA90p9mfnuLd6+bDslDVUAr9keGrbRYj0mzP 7PhB53IvNHEsBWi93PHlIAZ9K+lnz9McBjq2qep9Lwv1wJ1DFvLZ1pB6I0cwMvf6H+SARflesS3e gSEHkXvW42flO/VlIZmStfLwS3OIyzGHuLP2IZFOFN06V7qYSgj0Q0G+0h9EvsQyxZtDFOF4lC8j +rWfIGG6EqIpeb8jXVGFRjrbPH8++zWlY30l8ykNLLj2/I3oT9NPELT9fZj0p2fvpwFFH0ylv+v7 0+Ub9zg9v63ObO/fv7I7DSiGSpX3sfRjbWG2utrvY2fKPe5jp3Qaf8KQODGebS7Ru7l45T42V+XK mUJd5vn8rAwp97G5KncZDS3a9vz3o5HEwI27r1Ipa9i5/A6zhknLybhP3WlEc+z9r4fd6fb8z+I+ lrrLYWqchvpf0IzrUZS5lBbC4qdlqClTVxP9qjKKyMzrhbyQRfnmx5MgrdlPfd8khCzs9fK+N5fr 7sM87kZVahruvnaXvZ9qF8i1Up6Kj0kNVhbabmSRmuhnAPy8SA0rC613X/F3Ms8zi8sg8MuUmuhm ZKqcMtoY81Q5WAL7fXfDuwMA+nxtnhjA0qYhsw8hu6FG5jgi39MPCWlzVt3pSpcmppvhvpGN8cwd T/QPHFUqbfM/jBrjzDHvxDijJzg5tSzqi9t2Nh9l35sQhZrFPQghEpieJZ3S906p6ka0L8ovnnhh mV+3SKT2gbPCMF9M+t47q6gkDkRy0J8EIqHWTnu+793Rrx+GmCpd9L0tnb2rdFHBe5rxcWqy/hIR DWaIQ/m3Cl9T6w3Jw9ZdoknZ1MR/JQYkMWRjKbDRZc47BiSP51+M5yMPskh5+3jkQf8opyM1leCn 3SmJOA767bCvDez9XcrDQm81cardra/d5F/EtxLoOrlZx8+yBge+vmghWChzPjPtFcih7Xx/NAIW spqsv42AaTeUv30Ekat5PErI4anz2p8fGwED6usf5vEw+gKC6f+F/L7Bcf0L+wgIWC1MvePojE/t gxpozHtlpYyQK1th/tU+EtoPpPn57Z3xkXduXzZQnyEtU/1kDfz9AhJlR99dzd9PoIO9/37cOVez 898643mq39IZ3/SnxoNakLn+PshKmuJHtY/7UWec+ffL4SBYZvvrP2gwZcRw5h9AJUFu5ydAFV2Z M/uSXzPQnLyMrqpdqQP3/3ejryGsHdqfOj6GCHP7AvJPFn/ElT39wpRJU/tGZFBuHs76/y0699al PvX2wd3TT0uHcLCKaZvgsBwyUOmPg+Z5DI3eK2f/ZrHSXw6aBQ6du3/+UE6k3y90bf3b0Wi3zpXe K6eMdmM9etBF8xxth36Zf/T+bZgNcx+ogyxGv+mH99nxPYxm1bbjU8M76Riw2mriPY16Aei7G13J D+hQ9WuaP47PUzXXdq/6WTS3Mu3JbkAsn5Ud9Ld9aKdpGHvxfkq9um1PzGLRrBlMD78MZ0dzfj2P 7LYgo0p/LzNvpsgx0Z4qmTXB1DiwmE8UuabH48t3I2qm/fJpYQjz9SnzJlOtY2WR4f8AkrlNsgpl bmRzdHJlYW0KZW5kb2JqCjU4IDAgb2JqCjU0NzAKZW5kb2JqCjYwIDAgb2JqCjw8IC9UeXBlIC9Q YWdlIC9QYXJlbnQgNjEgMCBSIC9SZXNvdXJjZXMgNjIgMCBSIC9Db250ZW50cyA1OSAwIFIgL01l ZGlhQm94ClsgMCAwIDc5MiA2MTIgXSA+PgplbmRvYmoKNjIgMCBvYmoKPDwgL1Byb2NTZXQgWyAv UERGIC9UZXh0IF0gL0NvbG9yU3BhY2UgPDwgL0NzMSAzIDAgUiA+PiAvRm9udCA8PCAvRjEuMAo3 IDAgUiA+PiA+PgplbmRvYmoKNjQgMCBvYmoKPDwgL0xlbmd0aCA2MyAwIFIgL0ZpbHRlciAvRmxh dGVEZWNvZGUgPj4Kc3RyZWFtCnjaxVpLcxy3Eb7jV+CUoqssGN14H/2s+OBUHDGliy5rchluTGmp 3aUU//s0hhwAM+hFZCuUS7YoVn/zoaffjd138mf5ThpP/ymtNciAWjqr5WErX8m38qtvjyCvjhKm P8erDHXBT7AXjzhx85EcevpDHO+kVmgff8//MF7o/KR2SV69kd9cShcnKf2AqCXS/5dv5Fc/gNKk xeWNvIhfyMt/NzAEVNaieIJeyLUco/I2zVQXP63lIWRNSCyyeLMSG52Uj1geP6zlmBTo8CQXF1dr uYsKgy3P367lSasAXlb1xUJutVcJ6/m4et6apIyv/DrLRSMPoKL2S3nzvNOWXj8WuZvk31+Kn5/V t9o6zrl+pR1ZNpB3xbm3ByD3xKr9205uURlTvHvdyYNVIchzwQNA7knVuz+u5aiDgnr+953ckvVq 8F2uvAsYQYXq3R/WzxsIyrilfqKVG6Nc9e6L7nmHKsXu/USRewrf88kBls4P9fldIxeT3FD42vP2 t9609lnJBdjklDdFvl0/7yh7dfXPppN7rRyU+Ljv5Inev9pvv5Z7dMqkEh93ndzE1v67Tu5AGSjx cezkUasmu77s5Mkrn85mPwSqbnHgn+Ch5f/bOj6iNsrD+fMjUP7g+fiIOfur/J/d8yEqqPZ5uY6P RPHTVJ+vn6qL/MjqIgbVRZ6tLhicXVQX8VhdoCv90VN1KdrZkXbi99W+tXZkSTKJnn6CduQSTUqu yt/kvkeofILSX8oWMCWIWCMwUgj7CvlPT2KoimYvzpCTfNnzeA0KQgPa9kRk2CkdHzHiKd8WkOjI ZqbSHP+Xy8UnGJX0nQ6GIPK4QNXcr236Gq37yKgTn6aCxGQEp8J3j1ZqkNkh6LAofLE7Vox4xFhK berOFXPV0bicvSZUyMMMEQUSqcEhVMixY/HWqOytHlJYAsWGib7Rd9/RBCq1QLWqYKZiL1pIdBSH wVSI7FhiTFTQmlfaP0ZqA0kUhO1Lv+1IQJNh0LXniA5Clds3LHe73jIA5INYLCMuTgwk2dVRHQTR z0eJ2qU/vgqu41GM4lE/gvNPY3TugKlLCYN6MsgfV6GkhBirYAnKqXC7eb+VH7bysNkdt9dyc3cn T7f06/Zu+37z9iR3x+PD9viZlJzsZJ1j7TS7SnzyOFxUEGfsxKjw6nZzkqfD5nq7v7k5yg2dtrm/ P+zvD7vNafs57WMA/mz7MCr8df9BnvbyzebX7RRAV5vjVt7sD/TLnv5VLPeclmpTPTchsGHYhJ7B UssmxKrwqqtLlmZDACdm9EVf060PKqVYCJ/W30UPogU12lRZfu0h3lFnwMrS10iXgnKp0eVD36Zi rsWhsuz6ahysppMqy23fX7SjgdSOdKGOqmwIleXrDpLAKUw4skvuUrQeVZYt36Z0ddfTerOE0PyX QAzUBepSyjS6fMd0IE1rUmOX3YHBkJdo1RnpS4VAYfNKVwwkxryQFgjTMS3STh1HYQeWZtroR+al zZAGm8YwTPd2ZDs9dDX1d6eoKxfI6wsGk+cNqDSXPSTQGumaiGH0Dd4qi0NfR0yqsS4z10QaNmhd H710oo3fYWW57yCoaSmzzQvtGUhECvDRCyEg7Qt25GkEZ5Rzo+BF2hfyBUuB/KWHoDcqxCEL0hrd 2PZljzCBNhcYZSNa2vVjY9wrBkJlqk2Ahx7iPI3BMApd9JSNqQk6ppBRMtIE2yjzWw8J5EbfKPP6 iz86V4rfvefkprxqMdx0bCgvLFQ0E2r5AhRzBBBE8O61iJTqUFkOTKOiZclDZen96zAv+66y9Mkz LUshVpY9sywl5T1Wlrv3fRcyuWbbkTIhRRWe3vpsGwKF0LxSb95ImeHRVpa+3yWkeqzjSJeUjDIJ R4aZtqWAI5apDXkYvRFFasr3EIWFW7mQjIdhqAwmUK2vH5g25MjZqbJsuTYE0xo5sxyfcytbzYjR JHZE/DxTas5fRoMfWxuJNn2fwBc3ffbSRh8xFITsOdKk2gzomxptPQoaxLajcJhfJvKIkriRQnhG XHccOW9D8AOOAFZR1R1whJCUs9UcXz7jbeYizQEX/nq6zpSfKWAjFU0mXLbd9U4CYnPVk33qJk/V KFYvnDoOyPrPYm4epik11lD4pbuqAsgfSdSIZO91kPaEVBW97VkM7QkR+bieh+HYBvam55gWLByE PlgfuSPEPMDSoGArATMo59XKcCfMFN7kjaggXjDjrU95Sp4R/+o5pn42VCO3s3rIjtEj3+t54E6p g6vNJXtpjufflyHFwNZjg3guwcQnqFATTNRpG1kVjtxcPn2eWKuY6Gfu6RPLGcEMsahTEzMPPQWG acuYEcwEa8C0MUMDrOgH7qm9zpDfPqJYif/DxwC5t/kU16bc8J0r2hlcy71YjJ2UoDOiH/RspBWo yHf9vfmUng4LhJknp5XaF8ThGT9/X5zrE2cninqz8JR4tsQLOnIaSOZShyLJgZhN1E8hIdCG6Ksv H5hhOn8LJBaOvhYnKtbOVFddMc0LVPSucJyYSdrS7FojhvmUBDQtvHqkyDQlpxoR3BxN27mJA0Vo N6dNJg3iDjDkL7aIwSkGMNeaGXHLIIKnWlM4uCHbZpsVjrt7poUZpI1qpEjOIprjZwTTOzxmk40U 8eSaJttP3B1O/kxupEegndyHQYRAhMn5j98lYIyedO6TMwF3e+NDvlSZOZj7BU3m0riqF0tE0CqG wsHcdNB6pIIbRBgCDWpQOZgboryn6crxlruUCXnImjmYU/J9oqkcvzCI5Kg8jvSwXuch67xj0Sak zC4ct59pbM6dyLmuvh26kdfQoh8dzGDGYSZ5FYwtiHcdR77/8BR8M+KBvf5wFDgzYtcr4mivxoAD RTyt1ZSOBfGm4wjOKEM9/zwH5Yii6CkI5mPiSHOzrqec2Pt3VxU99qsIdYTo6iFMNmvt80XDjNgw qwiNMOCWiq4KsZm+7DAjrnuOfPOecKRHTiRTOX7qOXKaNPHxD6bMwvQ1thnxbc9h03T1MiP+zlRZ Kuax6vG+5/BUNqCegkyFpLfVbvm2S45A5cv6pefWuwSVrwK47ymic9S5BhEGCbVqovSq50ghqhqk uxuuzuar3grpSaYbcBuXEbS+3fb5uzmLQF5yECR/wa4cw1S4/BVUFwf5Ml1u2+oY1VS4/wLFjw3X CmVuZHN0cmVhbQplbmRvYmoKNjMgMCBvYmoKMjM3NQplbmRvYmoKNjUgMCBvYmoKPDwgL1R5cGUg L1BhZ2UgL1BhcmVudCA2MSAwIFIgL1Jlc291cmNlcyA2NiAwIFIgL0NvbnRlbnRzIDY0IDAgUiAv TWVkaWFCb3gKWyAwIDAgNzkyIDYxMiBdID4+CmVuZG9iago2NiAwIG9iago8PCAvUHJvY1NldCBb IC9QREYgL1RleHQgXSAvQ29sb3JTcGFjZSA8PCAvQ3MxIDMgMCBSID4+IC9Gb250IDw8IC9GMS4w CjcgMCBSID4+ID4+CmVuZG9iago2OCAwIG9iago8PCAvTGVuZ3RoIDY3IDAgUiAvRmlsdGVyIC9G bGF0ZURlY29kZSA+PgpzdHJlYW0KeNq9W1uTFbcRftevmMd1VSzrfnl0AFOkDMawiV/8sl4WsvGy wB5wnPz6dOscjfpM9wy7pNiyXdil77Skvnz6Whp/mH6ePkw+wd/aGGOn7MwUg5luLqZfpuvpuwc7 O53vJtv+2p0jNObUYN/ucer1LW2Y9hfY+DAZ7cL+v/FffFIGf2linc7fTn89nWJpo/CHLWZy8M/p 2+m7H6w2sIrT19NJ+WY6/ReBOet0CE4doCfTctwVnULtpk6eLsdzxpXAsMLhs8WwN1Wn4uaf3yzH XdXW5MO4OjlfjseiXQ7z7/+5HK9GZ5umsXx1NB5M0tWN+d3i98FX7dOwb3BckfFsdTHpeJz8PpoA 2y/zeGzjj07Vz181tiZEKbhpsTrwbIboqrXdWwvhKWP112w8OO39HN1XbDwHnfO0ljzWQnjqiO6T 5bgzWdsx/yM2HsB7I/lOF9G1rlidR3R/WP7e26x9PF6fouPe6zii+y37fXS6FrY/NY8nSN/14rAB 5s/j95dkXLVxD+kb1v0fkqf+WYwrG2rUyc/jF8vfR6heM+JzxsaT0dHO+fGejVfY//Dfu+V4clH7 OufHFRv3hfr/ko1Hq72d82PHxovRpLr+wsZr0qmuVr/NwG5lIz45WWr/2TI/ivE62fX5i4X6cev5 UbD6x/jf2e9z0Xb45+UyPyrkD2Gf7w/sMt2SXdQGu0yr7OJyDEfsovbsYhn1lwTswrhPXJ26G/ct VweeBJeY9qc1CVxqgbUX9LcPb4NOB6izoRHEAQ0ZqBgkRo102iFXghVIMyzDDrngVnyFSIYB2XEr oUCyDMTHd9xKAr7NA3LNjeQQdEqbSylAmsRDh6VsZ436f+Kyn1glC5PXwMLyC9sFJo7xeUYfiAWN dEgpWddKIDfMSg1Vm1AG5HdmBQ5K1AcDMj0WouvhDI+LmdQiAYDnhOXS+IMOSgPyiRsJtWqQIjPk PbeSjMVEG8t9zsOb4cwNW36Bf1pGk+UyI9WDMCB+ecOsOIPS6CgASyvOGqBnsqMLbsUZUBBkRztu xdWgTaSQr5ivNu8ntlmhQrY55mW+PtvvYiAnlzBD3Iw++alDVIf4GCBfy4D8g1kJHg4jrJEOmZiV kCKkSB4Qw6xEkAzW2gFxX1Fy0olTzaK7XiyXqHJB0Ux8ccF20TiqEo9+YFaqtRo7phnyiVmpGQ4g WOwMubyhGIV/AgVY7fzWYkAGA78Tl74dEHWAQDnobSMulVa7HXLNotvkpiE7+siteMiAQKzsuJXg POSRkEbDSoipUU2HPBGsFGzsBuQhtxJzU08b+QqEFalbdtxIAglL5/nEjWRQqbS6fuNWcvU6CAEa Rgq0GkepIDi3hqJp0n3ck/QRBHLXbwbamXDs3Fd7SqMQKFC9GSHQJdB5WLoUbiRHXclS3t0nLyaQ 5a3Qw1zoD9d58YDuByPlRWzwYb4Zcs2seOR/YuTiPglt3ucgtCec0GATBXT6vMRHnNCgk01+IF4y I6VAZQdi5LHIZ6D5B4QljoIjF+p2IF7zLEfCM1JQhhELZVsH4kagRNCZOW6uBJSItsOFXc8ccSLo WUOsfOJWPN7QEMcJxe/xEoZs6OrynNN8gMKtZE9nAp9B4Ray4I+c5mMGjUasXL4TGC0bvJTgGU0Y DSQlyS6JonPK2DrPkF9POBkV8A2k8VgNLx7gPWgRA00IxWgP0pcmnrCnCjXoiZVfv2lmbt9qrhah 2iSbdk1pQ2aq4nsWYRfzvggP6JPnnEl8uw6ZES9YgAM27JhLHcIjE6KDFEgDYlh8o8XMrwPi79KY qy+m5j1lwaRq4S3GNlOGwyOQTfDKK9AaeZDcM4RXXjUoR8uAvJX4yLVOokOkc9ekds0lO/2g0uCA t2S9TwRKchEE4YBc8/LF+8ZAUkRYi3OukUCHXPDqddB7ZmJFoEfvkk7CUojWq1aHKsxzrNJ8FlZ7 rNJy2UpXGx3o24F4KIu0sL2fBHRPCudMoDRTm7ztEOHgyXDwpM3tYO9qSLLsSW9BCHhnR8zsBF6E A8ES3wJb3aM08vEu0uiA3pJGHbIujTriXqXRvM8VVt5LI+AZZzc3gURjB+CMd3oGTssSBuTqP5yK ChzM1Fn/lTq93AiiQ6Tj0uKNNMG8EcofWj1HZhJqzuFBQ6w8Eurf+tbIdcif67pmXq+Q6sgRhswk MBpyBJ0Jtq14K7fXNWzbxxRgM9224son6Uhmei5wQGy3Ux1xw41gK4fXbawwSH1HaKU93bTisgfa vW0rFS92iZVz3mIZCFJOW1acyaC3yURXO6FTS07XrYRxFi8QSREA8TEreA9JK0lIX+fdXljuMUpg YUjN3G7cNvYU4Aw8ZLi6NX9u8Yq6pX5p/GkrU3t/Y4nvsE6xET+gD28nR+QIu4glDsgz5lN8/i0+ D4gg93xt6dgRhhnBN16L4rhD/D2dNo2FBW+tqD1kx75EfsvS5J4biBW1Z4jLhVsWVHvYtXTIqtpz C58fs1pTe2SmNbVH4iKQWlN7xMqa2iOhu+BWmtojVlbU3v45e3UtTe4lOpEs9yJdrpKoPKetdG1y z5KMfijQNOo9t7kjJHtHdnQm8DQKPrKWFcEX8+aOcvatLd/YEXa5YXNDSPYpbjoXyT4SKzuZ7AtZ 7blE9k4fOU682TNkKZe8jhzMpAcvKZHIXXX4jjfbefM1H09pi4QMbPJdFOwB3Z+Rjhj4oGA7ZF3B dsTFPd0rNO6c97nyuNNICV8rUvR0idJrRfTD4Mm/WetZ8UsOvCzvEJ7oNfk9v3bIc4FfD3d3HSKU rgEpVzINC7PSL+86RCDGfnm3YaVf3nXIuUCMxjStsWEFHzQcmehKeq5wXpMskq7CAkisQBazk5jR tF557JrL4Og18a5wR5hs0abQjOb0WkDtkeW+kogxNwYYEympzc1+M9IlWh0LnYhf7dnUjp0Nt1To uQvNS06MJkMPt5Uu7TE4HeWlkoRyyFvObY8elC4ERY7cmUgZ/S5QJ+gLWiP3eldZLZNjP/Gryoz3 XamDTx5wWjw8F3fEKbMRQHM4cHlHSBeV0JnkMYthNiLUVYStdIT/it8PLcmXOkpt6dZY7bzAT+Ll QcgBEWpdtvo4XPlWoExQrXHYWBOtJW94u2lWV2cjD2VO3U+zVtHQKaJ87jYE/sEvGt1YhyC68GIh DRvSGzIcV8OnAiUE6GhAOHcTEiNDYgCBdRvSrUOq2EN2G0L/HbGfDVsujSnh55fdhtAFxFr0SCbh 3a3dSeThjpfylcSokscCzwbMoC13FIgsqTQhKtUaPE833FEh101aZMf9CL5c76D3DuD9FpQk9zpi Ve11QFNS6n7E3rzHQczvudYDFeEgr/sCBb5xoIugvDriN2ajVJRFfkYsXiMH47jhiDP+VGFSxm9O OkKipFo1icelJPRqa+qPA3KXZyhdvozhMauST59/h+qvdgCWK1gdHu06gN8pBrxJDcMEf4qGUWCr MiOEr6YsnshuRoSlp9TnMvNWnlJSbqY+q/BR5NEXBECG2au+RE5U+NTrbZ038aPQYjhsMWYbD6Sv A/CbqXVnoq4qcBTONp5KB13CzqAjXkgfQ0Fibi3DQ4sS00bULV5S+rGMP6SLbwjqbMJJD/4oL9Tx Xu9CR3PQ1R3LI+Yyot5ZVwkRa5/8lpG6giTIIM9z36eSmCLjC8uwcfG5faovSu7jfapOA6G4TUU8 UUV8AO97SXa4QGs7I/g2PX6sX92M4PeKHiRaNWMWI2pmMkm48wG16qlbHFGCp17yIwqvco2flyhL Yldmc8IRhZLYQnfUESuSGDRPR1x+lE4og0LzyN1cEo9Znsrv8a7MCIkooDsFzdwRD7gNfxyy57JY HUH/g5sIIEvqmERiitj+Rw5hr/OXQXDsuxGWlyvfBQ2Xvuc22mdBw8aF9J3j0TrOuY32NkYCJ2rR hA+CM+ScCQdo+Q3ePnSI0KubFLHjZ+VIbIA2J4kqfCrlbIGjZED4ZxLORY+PBetJ5lxJ+FawsRBv C80hYR2+eGx42CzDRrARHwo64onwaBYKNjwdwZ98QTy3t9r1FIFT0eMXtR3x+M7P+V9KP0jUPqZl F/Ajv7qAc7mAIw5gITPwf+nMqcyIj8yGhyPNFjsj6Hn0P3SpDgEKZW5kc3RyZWFtCmVuZG9iago2 NyAwIG9iagozMDk1CmVuZG9iago2OSAwIG9iago8PCAvVHlwZSAvUGFnZSAvUGFyZW50IDYxIDAg UiAvUmVzb3VyY2VzIDcwIDAgUiAvQ29udGVudHMgNjggMCBSIC9NZWRpYUJveApbIDAgMCA3OTIg NjEyIF0gPj4KZW5kb2JqCjcwIDAgb2JqCjw8IC9Qcm9jU2V0IFsgL1BERiAvVGV4dCBdIC9Db2xv clNwYWNlIDw8IC9DczEgMyAwIFIgPj4gL0ZvbnQgPDwgL0Y0LjAKNTMgMCBSIC9GMS4wIDcgMCBS ID4+ID4+CmVuZG9iago3MiAwIG9iago8PCAvTGVuZ3RoIDcxIDAgUiAvRmlsdGVyIC9GbGF0ZURl Y29kZSA+PgpzdHJlYW0KeNq1WF1PXDcQffev8ONWahzP2B7bjy0iUaoSNclGeckLBaJSBQiB5vd3 fHfX9nruXS2tEKAFneNzx/Nx7Mu9fqfvtSP+NtZa0BGtDt7q71f6k77VL08eQF88aJi+Hi4KNUSa aC82PPXlSA07fbHGvbYG/ebv8osjZctKG7K+uNG/rnVIE8ofkKxG/lnf6JevwFiOYv1Fr9JPev13 R0NA4z2qLXWlRxyTIZ93UquzEY+xRMKwKvD5ADubDSWsy7+POGYDNm5xtboY8ZAMRl/X/zXi2ZoI pFv4ag/3lkzG9nwc1nuXjaOmbwuuOjyCSZb28W59sJ63nyoeJvx0rd49a22tD3PFpSE6zmzk6qql 3QNweVKL/lbgHo1ztbqXAo/exKiXmgeAy5Nbdd+MONpooD3/VOCes9eabz1UFzCBia26r8b1DqJx YT8+1ePOmdCq+0KsD2hyEvtTFSdu3+XhAM/Pj239dYerCXfcvn45/55cn58BV+BzMOQqfjWuDzy9 ttXnXOBkTYDaH98Ennn/LX93I04YjMu1P74K3KU+/9cCD2Ac1P54EHiyppuunwWeyVBenH6I7G7p QH0iQa//duyPZJ0hWH5+Ap4fXO6PVKa/4R/F+pgMtPx8GPsjc/907vPL1l30ke6iDriLXnQXjMHv uYvauAsI60/E7lKjo0PRqad53xgdZ5JTYqdPKhUFNu3B/f6Ynj8x9YaZAAxbRGXz8aNGSiwOh41y J1QymxzmjvJNqLDRksFwSIW9ko+yLpYHqYK0seMd5UqquHIg+Ua5lCo+OtOJ6DOpQuVc6MK9lioR 0mQ9O8pXqRL5cM7QhytUEhtMiv2mhUpmjw3djh7vhAyCDca7xrkVMliOEpfHYI6flcVuVAvdWItm PW/Ci3b82GJUu+JxCI28tdv92hXDpEa53FFUpXDohhu2Us5nysvHNqZGeZwaST1TMtz2yc5Nl04I XIYhF5/Rh+oO6n/fjGoISoaA2c2GsPHOnun48pAdqB17ZyAdxUduX2qCq/eCEoivMKFT0YJCfEnk Nm8qVlLYHJxPTSU80eprutQRZto9OLI9ynSp7VHZMxPfBYht8EAuckrlurRlqNWJYIBNgS/0+kDO p0slNpEfkoF+cpUqgpLiIJSTt1DUfFWmyx0frlXlk6R49skMTeW1pATuIuqy8rukUAjlllxVTv5r bbejoI6s7TSNEMLsND6jO47TOBPCb32W1DSNgd8UiSp755/9NHIeMx8VlfK2UdR2Grmm6LFRZN1D ymyO3YPqNKo6jXzt5jfDRnnSNKonXm32pnFTMTWk64OcRr4qIF+da4j/jBSVvTMuden6c2Yc+YLZ Z/RGqACU/wA0xvXMwJaX9Fbn1eOMSN6M/XJZAB2afkNnUsVNO2qU9zMTyy/01D3oRKoEx23U7WjG ggh9eW+olB9ShTIYRitlxoMi8dzDgU0rnlA0MRxIXTnQs4ldLHdSJYMrr2mHspsDv4l3Km9mVNi9 bdcLp0IFyzU3d9kVfan4/sXZbYzX3eT8Cy2g6EwKZW5kc3RyZWFtCmVuZG9iago3MSAwIG9iagox MTU0CmVuZG9iago3MyAwIG9iago8PCAvVHlwZSAvUGFnZSAvUGFyZW50IDYxIDAgUiAvUmVzb3Vy Y2VzIDc0IDAgUiAvQ29udGVudHMgNzIgMCBSIC9NZWRpYUJveApbIDAgMCA3OTIgNjEyIF0gPj4K ZW5kb2JqCjc0IDAgb2JqCjw8IC9Qcm9jU2V0IFsgL1BERiAvVGV4dCBdIC9Db2xvclNwYWNlIDw8 IC9DczEgMyAwIFIgPj4gL0ZvbnQgPDwgL0YxLjAKNyAwIFIgPj4gPj4KZW5kb2JqCjc2IDAgb2Jq Cjw8IC9MZW5ndGggNzUgMCBSIC9GaWx0ZXIgL0ZsYXRlRGVjb2RlID4+CnN0cmVhbQp42rWVTW8T QQyG7/MrzC1I1LXnyzNHQK3EAaSKRZyj0KotNJDm/0t4N2FmM8NG5YA20SZ6Xnu8/tod3MAOXNQP EhGDWILgCZ5v4Sts4fL9nmGzB56u/WaUBomT7OKgM3cv9EHTpT52QGj94f/4w0VDoyWFDJsneDdA SBPVGycCq9/hCS6vGUmjGO5glV7D8DiTWbbovTVH6QpabhNGn/+4Wn1sucgYiWIz4nWDHWWMyRbz 55bbjExy5Ga1aXlIaMUX+/uWZ0LhCDV8c8I9Rcy2nm8be+8yulj908jNjAtjonjKZ/aBvD5+KjxM /GowN/+1tuTD34obm+g0s6LVNUtPz6zlSTX6bce9RedKdb91XDyKwFLzMGt5cq3uh5ZbEuR6/lXH vWavNt/QVJdtYpRa3evW3rGgC6fxmTl3DkOt7kVnHyzm1D2fKTxq+y4PB3s9X6r9w4ybiTttX7+c fx/dPD8NN+xzwOgKv23tg04v1fqsOx4JA5f++NXxrM9f8/ez5dEGdLn0x4+OuzTP/0PHA6Pj0h/7 jifC2XS96XiOGPPi9LPodktn6iOR5/4/tf2RyGHk5fMT6/zY5f5I4/RX/qWzl4Rc8/O57Y+s/TPb Pm+P2wVeuF3Mme0Ci9vFSvAn28Uctgt3qz9F3S4lOjkXnfm33ddGp5nUlNB0H7eOjl727fobpgAm KRylNgXMuaj1/WFaiUsy9niRrHsvPmXMXCXb3kvI+qKQKvnee4lJl42rkn3vRSSgrYpXs4z+Biw4 uCsKZW5kc3RyZWFtCmVuZG9iago3NSAwIG9iago1NzkKZW5kb2JqCjc3IDAgb2JqCjw8IC9UeXBl IC9QYWdlIC9QYXJlbnQgNjEgMCBSIC9SZXNvdXJjZXMgNzggMCBSIC9Db250ZW50cyA3NiAwIFIg L01lZGlhQm94ClsgMCAwIDc5MiA2MTIgXSA+PgplbmRvYmoKNzggMCBvYmoKPDwgL1Byb2NTZXQg WyAvUERGIC9UZXh0IF0gL0NvbG9yU3BhY2UgPDwgL0NzMSAzIDAgUiA+PiAvRm9udCA8PCAvRjEu MAo3IDAgUiA+PiA+PgplbmRvYmoKNzkgMCBvYmoKPDwgL0xlbmd0aCA4MCAwIFIgL0xlbmd0aDEg MzYzNzIgL0ZpbHRlciAvRmxhdGVEZWNvZGUgPj4Kc3RyZWFtCnjatLx7YBvF9T86M7vS6q2VZL1f K+tt2ZJsSbblmHgdJ87DCXGIndgBE+dBEiA0toFAAiHmGeJQ4lLebUn4FgottDhPnEDB0ECBNiW0 lAItJe03paTgNm1TSgu27pmRnEe/9/7u9/5xvdbM7Ozs7syccz7nMSMhjBDSoUHEIXnVVSv6CGky QM1PEcLmVRuvkf5e+f7NUD6OkPDqmr61V/35goMPIKR+EiHFvWvXb1pDfrBlGULGRoSuGlx32YrV J9cdeBqhG5+HZ9SugwpLXPdjOD8N56F1V11z/bOD6vcQ2uKA8+fXb1i1Ap/2XwLnb8N571Urru9T PiY+gdBNWTiXvrLiqstMFwQWwXkHnFf0bbj6Gv4mpQ/OB+F8sG/gsr4T9fc+htBWODVaIMGIjoeO CGEl+n/9w4icKXO8YqqoFFTqYkmj1ekNRhGZiqdmZCmznvsAm93hdLk9Xp9fCpQHQ2EUicbiFYnK KvT/y1/y//MdirvhMx/54ePh7kVuhAq/g88J+Hw8Oa/wpeJKFJy8onCco7P3/dIHoTC6H+1CIXQK V6OX0Riah76DmlE7uhfNRm+iZ5ABbcI/QTwKopnoSRTGfpjIVmTHCvQQeg9dggbQH9BxFENt6LfY DM+ZhfqQDeULJyFtQ3cWDkErDWpBP0CH8Xq8GKWgPIdU4gS8eWdhDNlRrHC08C6cfQv9AYcKe9Ac KH0EdIgCsb8GhLgCvVH4knIVWomewDfikyiAetEOPssPFa5E09AB9EvcBqUFaJPiXfUBtB7u+ja2 47HCh4U/ohd4jC6DJ92C7oQe70VjJMm1KHYjCUXQBehCtAKu3oDewxZczcmFaGFG4SGofQL9jSTI q5wA/UiguWg5+ip6FGbjHXQC/QNrcQ5/Cz8Fx1v4z4p3oW9t6Fq0GWTrWzB7T6Cn0SFcjauJndhh tuwojjrh2k70OLx/HzqG23A3HsMvcY8r0pNNhbKCtfDHQgFVoC7o4S70ErzjNE5DG3gDV85dw/v4 axQ1EzfDCFejb6Jj6C3ox29h3v+BPscVcPyO3ES2FpYWniz8AfqiQn5UjxahZWgD2oiuQ/8FVH0Z HUF/xV8QNbR8k39FsVlxqnAPzG0EzYC+L4TWi+HZO4BKe9EoHO/AKE1YglHU4wvxRXgt3onvx6P4 PfweUZIA6Sd/4ka4n3C/4WsVikIDPMmGfPDeIFqK1gEFboLZvgfG+yR6Bb2OrTiCq2BE78D9n5Fp ZCYc3yZvkt9yt3M7+S8Vd0wen/xk8ovCEBKAy2bDPFyLvgez8Bdsgz7E8RX4avzf0PNhsp8zcCIX 5HJcM9fBdXN3cvdyr3E/4wf4p/j3FXMVKxRPCSsmvzL5VqGtcBsTeyX0K4oqURbVAf+sAW66EvrX B8cAuhHdjIbQ3cAv96Dd6CkY94vodfRL9AH6FCiAcAD6fDm8/Srgutvx3XA8hJ/GL+FX8Ov4d/gz epByOGKkljSRFtJK1pLb4biXHCPvkI85D7eK28oNwvEId5B7j0c8zxcUNXDMUexQPKH8iRAT5ggr VT/9cnyiYqJ74reTaNI1efHk/ZMvTf6xsKSwCfofRlUABjeibdDLh4AHH4fje8CJB9GrgN2/Yn39 GyZYARzvwEHghkqgWhOejefCsQAvgqMTjqV4GRwr8Eq8Do6teBDfgm/Ft+Gv4vvY8SCM7XH8XXwQ jmfxYTh+iT/EH+E/4b8RYGLCATeHSZSkSB5G2kJmk4XkIjjWkg1w9JEBshEo9ATZRw6RdzgLF+aq uBVcP/cQ9wPuZe5t7l884Sv5FN/IL+HX8rfyb/Jv8e/yXyj8ilmKdYpHFC8r3cqsslN5hfJB5TPK j5VfCkqhXVgp3Ci8LRRUYUCrH8O4D5wHeSnlm/hqRRl/PfkQ5MLB9Sm24U6YMSXp4NZzd3M/V6zB pzgJv4+HuMu5Kwvf5lrJ59wGvIS8iMs5v6KBW4PuQgX8FPkdOU3+yFtxBzmJY/zX8LNkA9dCmF5R /IK38rcqPkaI/Ao1kC14jLzC3crdWvghalA8gj9UPELeQhJ/nFjQhyDV2whoTfQzcjnZgbr4rOIL dDnM+3cV18N8Tyd34grubf4R9AcuSP6OT+H7ATWO4nl8iFxK8vgpQNwJ7EPjuB/14fuQjJ/DH+BR hPGT3BN4PtEBtUaIHteB6jvKBfDbnAZ1M+0WIVbcTk6RTu555TEuhzGgxM/RZszhNPDO1N8k+gpI wL0kCpg2C9DkF7gGOdADgPenJ5+niK14V7ED+OxRrhJdhNKoh/wENYBs/AGOLnQHqkGHgQfvRGny ILqxMIhXA+4vAPwkaBRfgVJYC2hpx1RJ88hGygELl8NbPwf8fwNQvw3/GV2HJZCsMRTj6ZW7+FmA TL2AvzvgWI164Oyb6B7lAcUv0EJsR4iXJh8BLv8NuhR0zn/D+12oEfq3DD3KV0KvJUDmfrjjm5Nz kAzHHegnmKAt0OfpIOft/BxA3vsLV8AILwcdNR904uvo8sIDqAVod1Hh1sIOtLzwaOEStBYtLjwJ +LuxsBfVom2KbrJEkeCzgLGv4yOgj36NdwBuz0HvAx6FsQP9CY4fQP+nK55DQ/yvADubCncVfoms MB/lMEMrQYueQFehP8O8zeHGUGbyQrKn0Mr1gYb6EC0qPFHwYw1aV1gPyPs8elxQAPYMIp/iceBd JM/o7JCbpl/QOK0hX19Xm8tmaqrTqWRVZaIiHotGwqFgeUDy+7wet8vpsNvKLGaTaDTodVqNWiUo FTxHMKqcFWztlUYivSN8JDhnThU9D66AihXnVPSOSFDVen6bEamXNZPObylDyzX/0VIutpTPtMSi 1IgaqyqlWUFp5OjMoDSKly3qgvJXZwa7pZFxVl7AysOsrIdyIAA3SLMc62ZKI7hXmjXSunHd0Kze mfC4PVpNS7DlMk1VJdqj0UJRC6URe7BvD7ZPx6xA7LMa9hCk0kOnRlzBmbNGnMGZtAcjXHjWitUj 7Yu6Zs10BwLdVZUjuGVVcOUICs4YMSZYE9TCXjOibBkR2Guky+lo0A5pT+XY0F2jIlrZm9CtDq5e cUnXCLeim77DlID3zhyxbz7hOHsKDze3dG0796qbG5rluFyip0ND26SR3Yu6zr0aoGl3NzxjhIRb e4da4cV3wRS2LZbgXeT27q4RfDu8UKLjoGMqju6y4Cxa03uFNKIOzgiuG7qiFwjjGhpBF20K7HW5 5EOF48g1Sxrq6AoGRprcwe4VMz17ytDQRZv2OWXJef6Vqso9oqk4rXsMxlJBpz+3cNmZa6zEmtNS 20Vn5hXTHgXnAjuMSKsk6ElXEMZUT5PL6tHQqnpoBn/dGO4aWQ30uHxE3dI7JDZAvUjvH1GExaA0 9A8E9A+Of3p+zYpSjTIs/gPRIuWSM4wG16fKI4nESEUFZRChBSgKfZzOznNVlRtHyUiwT5Qgg+lD 7TC3K7obUjD5gQAl745RGa2Ek5HBRV3FcwmtdO9FcirRPUJ66ZWxqSvWTnplcOrKmdt7g8DH+5lP Yh1RRc78G0WbZda6hhFs+z9cvqx4vW1xsG3Rsi5p1lBvaW7bOs47K16vP3OtVMLFCzDhI3wYZmpu EFjvomVdtAL+FeHW4KzLe+eAqEEfRywtXZybdBdLxM2xRwH/XnLmyfSkS0efxYeVjP9XjwoqYGBW g6XWEbF3TjHt1gQC/8ubRgun6F0sO3tbaUwjDYnzz6edd35e93RDHHSYj5C2jmVDQ5rzrrUCWA0N tQal1qHeoRWjhcGVQUkMDh3iuriuob5ZvVPkHy0c3uEeab2rGwaxDjcAaxM0gwpjS0fXudPMeLe7 ChFKOQUc4D0LaMEegp8jL4BtKZAX9yIFP0pe2M8hjUALBzByqpSKF+E6QRyOIzW+El+KHAnxs8aJ xgvF040LJhpRE5TFLyGpTgdMAVMYEgxa80uJG/tSVqAvwKIYg1cSdDtYCs+D76YH3ffNZ0edrzn/ qeN0o4XP9wXDWZZXpbN4tPDxvopcFo0WXpO9UHA6IHHVQ/JPHRZ0dh3ReG43rK3Vg6bu2CdwLgPk e8s4NMrl9uv1Gt4ABdnmctlNmqv4H9mvQiZsut3tuTdwxWZHIvFZz8Rn4yZzPlVMUNNEYxPtegL3 9yTYHx7AXDSSy9ZmamzWMoELcOecELnWRuqTibwlP7myzgYi2eCq5YI4tMnpbGpoqO5cNflrHNtc KTdMq47ePfketdu/MvkUfhC9Bt7TYjnaTbrtR2yc2t7rPObk1BgJPG9UmdFBs6zT8g1Gq986aOWs o7hC1vqNy43E6HR88zFHAma7Z8FED3R4/IQ5j01me746jXtwvyVXC5o1GgmWC8pg+ZmuKr+ytl8t CNqwuay6oa12xtqdk09Vlu9st+jVZeqGTHXr1cvX7imGExbjQdIF/h2HmmSJKAa9q2u3KjANM4xw HCIibgecGsa78TGsBMcpewAN8h3LKBdM9DSKjSg1DintTMISsAYWE8XEF8T+QPHZXyucwBvAFtKi hOxBslLLyWq5IaeWm3LL1XiX+hk1Ud+uo5QRP+sfSCTo+KrTYTaC4mgwSsnNyWRz88ssTaZk+tz5 hRPgc86H5750CPGF4/ss7umK0cJxOQEFpworuAr1DCTre/W79W/g18m7+F1yXK91IfB4kV7Wc0TB 86P467KLI2UcR3hOr5Bn5xS/x0rIlL+nHDyKHzq4W4u1Tp3iMPkYceSPsg7xIi/z7fxuXsE/Tz5C OuyDmaK97xk/wWh0GgRiPCGOJ5oatymSiW2GLUf2KKkwyuprFNcob1PcpuRRTzfMV/9ADxrACTpr 4O0EgH5C9GfkV5ONYDBP7uhPd2S8ivmRf7/Av+JO9mpZuAds1Y+5IcU65AR/N4M3y4e7MVZn/JmK 6IbM5vJB7aBu0DXoviU8GBnKfNfxuOuJ8D7dftezkeeir2he0f5KbxOQBiv1xKWO2vR2V1gfNrTh u/Ct+tsN30WGaagBt4F1Oze2HF8cvSRzBboCX07WRq6IrsvcgG+Mbqy8MbOT36kYFAZVt5huMe8s 22l7kL9fda/pfvM3bN+JfD/6/cwof1B1Uvsn3UnDyejJmrigV0cbUB7X1yhmqpDOFeVZItrBpFLu VSqqaGbRe5vV2Aj4IrNPGsoiXiOLKCfniJzrze3OHc/xueDzcIHD01EFNsqatF22D9s5uzN7GP8Z n0QlIRk/PS729H82fuI0ZaamcSoqmEpLyya5JpHylZtsvMoaDiiCK5Ff8K7ElWUVK1HSnF6Jy3lp JfapIEnYqlailAkSxDCBIcPN8Acwgfop3SJnRU2w2Yv8ymAjTOURxFFpLbPZLUqaZWqYiOLtj/b8 9LuPvbb+qZH8/Pf3vLR+ySZcfb28cc2awVx17eL2r161/pbIbPLUbbuX3Pbi3oH5j1x554Vr+nf+ ZNOKq5fteWf9loWXX7dxYXZdavKPrY/33vyNzUvn5K8AKV0EsvAk8IQdRbFOztwQfU/xq/L3ovw6 fpNii2qz+jrd9fpNluukHapbLWBZ74yTaSpF1BGIOhScL8wjQXEYr0IOLO+PtgtYGMUtsjoV3hAm 4TDyUfIYFCD2d+2325HecRhm34WNzyKzaJbMnHkUXyabUVyOD8Y5Od4b3x0/Hufj4HPLKADNZM2L GqJxxg6DnHx8hkInKIF6Jk6AqIhAIQDinvEB8TSFZEYtyIv0qnCHVCZdRAx7IsGIXx9YibxG10oc UkFJ0vpWYrcJknJ1uEinIpkooSiZeuwAjea6Io7UAa0olQigJKYEKlLITmmz/pbjb8W/tXXnT9fc 8OoT193z21cffYFkzDM2Lei+o7t5efImT5hci0PPXPbBs3t3fHfoqS9+P7np5ivIoVsuXPG763c/ 8ovrllQy/TYCWDfMjQAm2dGMPZxzFIdlr35t7bBzt5MoZSToZLPWKFsBArPD1t1WYn0eh8Fr/Tlo ZIYgwLhFlmVgis8BQcu5gBigMAifylTzDJpzI0VkTDZPWGYUSzNof0TQ8yOgbz3ITxx7CMMgM/b7 iM+LPD4P8vqxz0PKXuB+j+zwEeCj4X4v21XE4+OMKo/Ni/x9eBATjFVGokKpJkqoo8eOplKUSuL4 +J8/xanin7hl25EjInyq027ZrTIYjXpR41P72wNKq9Eiukwut9vj8CoDo4WxveEczfalu7IsTyRZ vjderJYixWqXr1htZ9V7rSyTHxAtWb1RCw/PG+cZW8W5voWBbuNSsbOsy3eFca24zrdRHOS3GYaM 28Rt5u2+O/3fMH5DfMj0Dd8h4yHxh65Dvp8Y3xBf877h+7XxXfET48fix75/GT8X/+X9l69SbWxz Ez8gOkwS8vp8HrVB41bbPHa3TUUEt8pqKnNbr/cZRUn0eTzlJrHM1AfWBTithlHyumwivjJCfH7v 4wgVJ24UH5B1KtHIWW02lUqt8ozif8tqI9xDHjfIplGS3rfQh32j5FPZIMmGdsMpA2d4QrpyiPGD 0wUK3+ESAc/GqajAAelpULoTjdsMyYRii3hkW48h6UhsU2w5knAgcRyLY/8z3SZuOdIoNMI/1Tol M4dZOj3dOCAwoLIGcrV1tXU4g23FE4pYWsJ9d+Lvl5RPWznZ2enMTMcfBPG7+Z7FEycX5WNf+ehT /Oo7C6P+lBAOGx3pr/OXfPHgnYsU4TCfDFQux3oSmvgN1VnlCPEfga72oQSqJ1vk9DK0zLcd3enb nnnI9a3o066noyddf4r+MaWrR5ujmzIP1zyUeTz0vcy7rnej78Y0fMMo+eM+49raBsoVnvIszeX/ ttqzGTlQCYnTl62RgzFI3N7szNDM8HbXe/id0PuZP4QFPoTD+hqRsyrdrjKfLWSLWdPJmlmhedml uMu5LHo/MYlIbOjEy0K9DX0Ngw27G1SutKumHXGi4Ar5Ys4UryScz+5bmLkz9HDovYwgNcgN7Q2r yCquV9Gr7BV60xuVV7uudvf5rgldHd0cu015h/sO387MYMMbqfdTn4T+HXJ2q4x+tzpQLvrdtkAw E0IcX4lyCX+IK4/XV2a4ZHksl1Pb4jG73UaSMcopwxEcoWzfkGPZDJoN7mtqztLTfS2tLJfLoH7+ cg/W+NIe4unkE/76ymp6QZyVM8tgpRAEyXGe42mlRm/KIh5LPAbT5y05XKm0WEhnpc5opKleD2k5 8LJRJJ1GiZ4aH8k3PI/fQgG0AjsAoxIXnk4kGheMA+9M9PQnevpbug6haq7qpJtl490JsbGRcujA OGOwgSLAw4cZ2wAaFDhMRfM1kWAQ35zKBmMOHxZcbqebKJWRECieTCTmiGRwSqjO4KAvkuGyuDrD Rd3xDE4rkhkU9pZnkK+Gy2XA+hIbE43ngH+igunpnn48MDCABvrPKHAEKgEXVbUyGMhlaupqcyZq OQdzAdANtD5so/qgqL8FU1FvM00hcHu/2rpi8MM/TAxmOsN2b3RBhsx7bNX9j9w4cUN4ef6er1/4 8uHV7df0H3hhycs7p3e5yX7fjEtuv+xQZ7g2OMCtvylQGXaEnr1uzaNGQWi6ZcF1T9q+2OD+9vUL 7+mgK3gYzSv8TmEErA5hIs9Q+1I4RVJcyn+/8SHft43fNh80PmvWqnzQe7yFu8F6ve2r3JDtW9z9 rqe55zi1jjPwxDuH6+YUKZVoCrnBdFUcIG6MD4NX1HZQelgR83B4lHx4wJQYEbE4yjUf2KnfpSf6 US4lp8rU5GmEMa4Rn37GhP2mJhMxuWRgQHWj5MBGh99BHIw9HHPDq1cxJZ7oGVgwfqHY89lAPxhc /dSP6j/dc/qjpvFPTwPkUBvsdUZeyepW6oSwK6KN2MJKt7oK6ayQqJyKKqyx66uo3sbnau2B/h5s CbJJJ9YyM6VBnV3JByVqXJlDVItTytXxb/n90z96dNv7WzaOP3jbG5v8ayZPPTf5zKGhg7jph1/f WWF2l7m0iisnM28e3D759oejk38b7n+y7MCT/z785U9wx3NzbBZ3GjG9HQQ9uQnwyQYWCyd3a91a 7x3ifeIvRcVGcWPZNvFBy0PW192ve98WVQ6Tuczr4wQr3ua600diKqXfjcBy97v1gaA94PTHDAY9 ccZsNqTyNC4046KZlDbLZoV5tPDbg3QWzXODVBqnN+XkIJaCuC+4O3g8yAUDdiaPdiaPdjbhdnAK dCLIo5JVKl20UvlI+YoSFag0TrAULKqBxGeMLGeFLj8lZB6Xz2gVw2URn9GzBLuskHhN/iXYbXEu mSLAzTcjKjM9/ZnzRUPizVZRUAaiMO8I0BIkI5hZErJ5qAzEcBpf8NLTL01e++utSz7GNZM/O7Xs 6nBd4Gpu/VapMjw0+cIvJv/wwtsrPbgV27ETz/QW/cIK0An7Yc4zuFZuknNrPdd5vpH+ruPp9HPp 4znVEmefsk/YqtqqHlQOCjtVO9XqkN/tDZSH/e5EIKiS6ZSoAgaDX+1WCXQyA7RGCBDiV7oFj+gm OAg2iDeDHk8kUZVYRapGyS9AXVQmgKke97o/9ni8KvXTKpXy6SZhq0CQIAoLBQ6e9ZHczp61Mfl0 ZcJflYJb17uelsCq+dDNuRe35/rAGeFySGTEEhldREYssTwcYsQKscoQI1bokezxQ3gbM/AooRi1 QG56xk/3nJgAgvWA98xI9ilodcgmmXoHuGycaKSGsDj+KRL/kcClnBIUYKwHmwJUCjKmYCQKwhIw lQF0UVmBOq4IbmdJSOUJSvhpXHFNNKsMhw0G80Wdk++IsfqPrl6Xnt4cu/aLT9LphGR3hTrSvNUY tWZqYpcpyMTHweQ1k7FVnmBssnlZ1C6lpm+ZfDpsF+VVXP/Nvlh48ldXtluNRZoGgKZ0zb4KV+yJ pUaxT64Lr65V82rNSIp7MHE48WriPe4XiZP8Sc0X/BcadZ+iT7kVqDyoGFTuBCqrBI26gggBnW4U R2S9yi14/W57oFwJZKU1cYVbaWAa1Od3RwLBRGVMo9LxCgLEBgLYq1AwgmJijMQorcPRaITY7Kpo IvY0imMUT4OT0ge+ybBS6RfwQgG/yJydA3ISGRgtDYxsBkZLQ7nPy2jpZZVeRkvvI8n/IXinQe4a wVbrZ94M0O/PPWfIx9wZ5s0kSvSbmMqBiIBzoEOxiRINyJgkwaAJ/BGAt4z1HO00RUG4jr/9z86F +nAYR2fN/KdeI1WmqycOpzsiDr3GD2zB/VUfdM267Aog2ydtGyZzC+eFJ5esDTjNjnC4WtrMrS+W J99Z3h2j9JoDOud7oHOyuEfu0PCtSeKMumJEdIhOItXKtb2116v6HH3O6yuGHcPOEceIU1uV2qjd puUctUlXe21f7V389/njtbyOu0M7VsvNUQFdHH8vN1OqBbNMC+1jWgjvAzuwTW6pfrjS7nCUK2OV nCFWrsYJv09HZ97HJtmnpJMM/rmp3TxsJkbzQjOh+LnVXDDzZp5SwwwgemI/A9FR8rms1TS2R7Ax 4o8QMIdOySJ9TESk1yNzc6uHSrQCUARJSyUYqRjVTjA3i1JJnNJXJaTMSglBVIVj0Xi0IsopdWCO GAOmaVjyiyYhoalC+iAkomSYhtRRZRXWhg1VJdeT2ufU/CgqsgSTUmqAUHUGVJSooV3UZyZqVOQC Vhqzs5rAGmHKDUSXBhSYS1rHnwSyd2x6YXJiW//9fx9su6vZ33wR0Tsv9JZdfXz75HU/fWjJmr33 /WTepg31FoubA0XXsXvRtUe//5eXJ8fui4TxnWuaApFINnzV5IrpDV/+8J/7HvvR5UsdcWsww2QV UnI93WODXpI3BBieBmQ6bwE5lnMGVphW16r8bhIod/jd5kC50+/GgaDa7zYFgmYTCJzK4SSUck4V nXInT291lqv7VIOq4yquoMJpVbuqV8UtV42pjqk4FU+bqZgUqUYLn++n90JhUvYyKF8h9QUGA8cD XDrQHugNcGOBYwGy4jdAP6AZEzcgHlCvKHNM0BIMCmka/p/iUprZojiR6yeeK0lJZTpNZlUvjjhB ehLp8HlyQctf3svKRbuA+i0JNkeV+PpDKAlq5usNuVTyWsc17ms8N8b6kvd5hE2OZ0OHY792/9rz fkjpjIrJWCQfzkenxdLJZdHLo33JwaT2VYRdnrinzfMr56/diidj+I3Qe/b3Q++Bd/NJSOmRg96Y CoBNFSjHfrcQCIKzYA0EkVeqrPDGmoILgwANgrUCrAorUQkqM3KJ4KPIrj6XwjU3WbIlUBLLyZEk 2ZUcSx5LcslKzEANM9HCDNRwudHAQK2EdIwchkeqkqP4un0BCm3MwP8Pm6JnAbXyI0UrP0Kt/CLQ MZu+B4QoT/GtZGeE4naPIxyLxO1gvoc8kESdFRkcdgPfnbUz5nZskkVfecAfnMaX+6RpKCD5EWby gxLFENsAiE4PjYJQIp6PiMxet5UCIlHbWStdwI95IguyE89lloTL3GCe4L8e/Pnwr1+rHmjOXeRd 98Cc2zoy7eSGyWsH/ZXhcL3/Gm49LbXt3fydY4bZGs2jg10PtFnO2ITrgPYxlCVE3hty0AkLs2nb Vo7Nt0deCb5Sxc0NPVFFHH57ck2IU2N1OBKejbrwBrIhdAO+gVztv1raWH59eAhvkx6sego/FX42 8nxVIWRVSrfhu0K3RR8OPY4fI98JPVP1YtW76b9UFar0ZmTDLmKOAX2rG5IN6TWhy1OaChXxeLDV 7zYGylE45kYgmoZA0OZ3ewJBmVSGQ6FygstALENPE4kIFfHHmVFkp90Fw6Zd6BW4YWG3QATkftqT HcVfk401Ma/XQ4wGA3hQKjMLsXQVQyyzFuZQ4JkAWRjYHSCBA2ItlgHrj9VytVkV4ykVmwcV4ylV uc3KeMrKKq2Mp6yP5FYcwk70H7pS7BkAdZlIUH5KFfkpVeKnUmxjfFwEhuoZSCUmoMLpEse30fDG FvEINuddwJ8snpHYJiq2HKlOOyjHVVX7gv5wVTCVwdU+SJLllRkUDKWlmgxGgMmM6QCRB4q+Rcsl 8PJw4fheXR6Dythblo+NFo4fLMsT0UmLpw6I+bRozKMis6JENzBhIhDAjNn+T8wo0MgJrimxI3Cj Yt3k/ZO5jKT3iZ7I/BxjSyu1mvGf3z2689tPYUfv0IYvL7B41C+/suvWhlVkM8F4cuP5zNn03Wu3 jEYmb7ijS0fuxU/esnWXheL3YOF3vAJ0dz1ZKjvN91ViIzYSLYeMfAzFFYmFeCFRmxpGcat8rLa+ 1sW5+eWO5c7lruVupUKvMKCKsQb+Gu01+msMG419vj5/X6ovvV11h3abfpvhNuO2xJP8kxnRrM/o s/qcN+PNenPUJa3iJZ/kj8erMtPxdNLEp51pX9qfDlyQvSA3Rz+nokO7RL9UXBJfkvD6sZ+4M/6c u7bD0eHscHXXXJK5JHtJ7pLaZXUGTquNW7TueFArNUyLpxsGzAOW7aEHhQdTD6WfTI3FXqp4NTHW cKqh7EJVvRttIO5n8JuY4K245NHK+tzD1R63d4Pf7fMd9tKarPPhsgrgMZ2hTKczJHQVBj6iZpky iCcQUsaquWCMerpY9pVnMfbTAAsOymLK9KKJfGjCkukZ04cmzjRKtj3rf9qXEEGiaQP/riR+MfmX ZAFAVZ6dk5NvwgmHklIyDVDLJ5/HrSgP7o2jFN7uSfSDEzZwenwC4HNiIJ9KFIO6DDHtxdgHjd4Z gKsd6Iz9wUo9WOyHMsPS2lBasMQi2kp1BsWNFE4tkAhpONVU6TJIq6tMREUAV6MhXhE2A8CqUkrK 80VDhCVT6xYYeL+npUtWr9Ku0a8VVyX4nu4eDOiO+hEVCVmndRjzfNqYz6Qp7yeA7ZlJWoyR232k tL7BYiFBU8ZHimwejYRKCyFFs4V7KmzuefqSdXcmpp98YUfbX56flvX/yOX0CuGwq+vA+i1fq2uI Tj729fnHv79+U73dFdCA4ZLYtvvSrYumZ9q2rLnq3kUPf6hWNPlS+K17vtZ727KaNZW+H11zV8c9 v8g5/aminzEd9PII08t/lRuW4WVkmXeZ70p8JbnSe6VPlQo0BRYGHlQ84H5S8R23QLDXB0ApBsrV FD+DgiOI/EQ0qgKjZEy2gA2KZLuhyWyEx7WjZ+iyH4nJLpWaIZ2agZqaIZ263G7zJ3wUIQ30DuQT fct9u3287zCJIVvhU1lLcdDGENAGT98nre4pun2neyjk+QBitTn6gL1aYxamOHFCbCw5EZQ2SNbm 4DN16SOmZicaqe/wOo2mYBq/shT9veB/IBG1d4AwFv5RY0Rr8a/teNEdWZiaeCkNjvq3l8ey84SI qJg/+XJHqKHui9Nb/BXhcFYa4HUGy/pL8PSivTO3MM5t555BNegCbm5pvUBqYlZhk0xnw+oWkmGV Vks1Ia0NI12Gmt1as5l0Zmy0CZz/dj+dBiiclq109jKsbSYvsFyoYgaLpIZbkhnk4+OV6axOVsND dbLXS1MTXNKNFt6WfbSRTsdvdWAHq3WwFg4x7BMaK3mUAqk6ApYh2CGUzY+mJujC0duJozgFJ4z1 x8Y+SCSOiG8fpaaiW96g9QxliHlxLTZL/vxg05PqgxrOnDBvQVsyd6Ad2h05pddsaxCbBpt4tWe+ Yr5yljSrfH6D3LTdq9IYBAmVz8Vtmrnaubm2upaGuRcs1a7V3q6+TXOb1thhu9VG/E3Lm0ivKoOy jcl4VfY57EY6pCuMHVTndTFtXkfH7mrIibp2HZEh6dVxEss26nhdo2O08K4c1+YXOpY7Nji4lGOr gzhu8ouYjjjdKDcSGHZf1WAVqcrBvI1yrbKJ1ybHqnBVbxhl9DpdNgsT/yVQQNmZeQ6vRSHQdfBG Qx6F/eHB8HCYl8OnwmQwjMMibRR+jrQgAVmBJ/156yheK/vcqXy1IBvyEtgNgwInCviUgOnCYMv0 lq8UIa5/YCBB11kT4E1RGxG4mGlwkXrCYJ6fnjjRI473N40P0DCxKU/bJBKp4jr4Xk6HUU93cbGv tM43OzfNE1RY6upr64lSrdKoiDJQLpUTZU6bl5DJa/Egs8Xo13tweXCaIu9B9aqshHNZrdkjerCh HJIGZaMHMSuDwh4DQPDFKmgwGA/gftQPcAdY17W3yYx7qFJHAwB8+6thpMCRx/eKLDtoyNdJMPbR wsdgIUh0R4FWm3dI2rwdPh7K7S5tXgOkrIvRXAO5BnI15GoGnOf+dcM4w0phagWytq5oGCit9rIz ezWoi2Jl0Rsaz7EWzQy4p7huTGZ/NVR7wfIbfPGffLp0cVM4QlKRcGpk1+YLp3nMGrtR1Fkb+9ZU N+AHKhfOXFI//7arTM5brmipnnn9ktD2NeXllQ3JmmzVkuG4f0bi9snXb51WJugb6++f+XXc0+is 7M3PWc5kv/BF4QR3SHE3sqEQ/nlR9vf4FFSGRSrNijIdcmhYCBhY+CPmvukoo9EqVqCSrqPt9bS9 TuewI56oLaPc72VTmayGZmVW5A6rtYFusEKpPmz6IFFUiExSP0iMia+C2Fan3aXtEhF4BAePgPvo PfRen0IRCSMai1Z2OgjlX9qdz/fTcyj8+VlapdNFwiYGCSD6Y7R0tPS+o8VdM255kxjBjykPKg8I f/LzikiLvqdWilzLbeTv4Lbx3+GeUgmzBdygKovqmy2+spkOuw7xbhsSA/hMT6r9imEF6VUMKp5R cIpPdDaEHCGdTtS36/v0w3p+EJIRPYf0ol7Sp6E4pj+mF/Qg/8825vS94ZfbSnF06tWK1NoVJ3oG ipbCQJPJnv/H+Jf4H0w4Yk6J0woRifNJ2KVxeJDTodV5VHDm5wMSdmrdHuRVuqXSEngpDAFuVj9m UYiB7m5c2r9U5K6isxQNZ0wm21lzVYmn3f7wV3/+Xzuean98iVFyeCoM2FKVuSp/8be+tTqXi5HP Dv31rdP3DTY0cAe+OcclBvsmYhO/qcm89uLID91loJdbgYfmgf4I4H/sVfF4SoMQ13khbKYFlLaw US30BvrAwYApOUD5KeAFzN9vKSOdUHjjINUp3moOQB4APNHTdGScMcpRusK8x8wi6FdXVGVRkFLP rl+qIB5LB79YsVjZIXS5uzzCWsVGxSAaDOx3vyIdk46jPyjUdXg2XuLo9CwP9jp6PRsdA54h892W YdOw4zvgiD0T3Idfwj8Wfuw8qTrh+ZN0GjuUZJ55qXmHf4c0GDwVFEwSfr5wHEnw8QNkIC+iEJwG vugNDAYICogBiQUw+gLD4DyNBGgc43jgVEAfWOP9EEz0H9vCagGG9y74HTST6815GKQ28FO/Di/U 7dQRXUpEaSSjXtSHhtEIGkPHkZpWEPS9q123uki7C+9yYdco1snmU0qMlKJSUqaVslKhbClvOUS+ VjRD6bpMz0D/RH/PiX7GVolE0/h4PwPvE+aSiGkWe1d5r/ZyX/cCIvd3g2zU19fjerZ5goauwFUD iESiI+8G5DtoyStEkbpNY4CWgI1je8R8aXUNWKwfK4G9SC6LMjVT22EixX1pDMoA3bh54Xdv/ebH GO/f9oPqymk+kzYYnL76gkWPbl95YV0WX3LgR1j54bvYsHNBJBWxbvT75q189LEvWpKbmH0ys3AC fJ67wVCrIm0l7oqkWOwqrnQwtlIVWYyxG5K8NgZZNq1EgclEOUpi8UaJtYbaz2XGlJKD3iF5DnO/ R16qrGng12+m4CVaZLWBdFrKUBhIV1nJMauDYlcKPrhkZXwANsYYY0+wM6YA7CIz3IUkLcfRWz19 Xix7e73E69fCY7Q2hmI2nkIW9LCM5hJvNEJK6BVJSiXjrA0bnLJTqUwlGa4dTRThLTF2FBxU2pme nqNNdHUVIA6kA7zqwti+2bOzKSokMxLJbG/qRv5GxRA/mHomNZYS5NRgiqCUrcKa6FR0qjoS9wvC HAFLqTrNbM0SzYP8ExW7U8JY6lSCSBKSAoeB37WgCWc1SgulS6U1mvXSZmkX2iV9TzgkvFqhjags UV2z2WeZafVGbc0en3emH27T8pVWNmv+SlxZ6ee0fqQN6CRqZJitvbZB2zM2zm8bthHbJ/F2JV3W jiWzNH92dk7ZkmzZWgofLBifGOgBQ5j+0c11AzBkAEiRISQSzwKlK5LgVdFwRBWXUIKHJCaEJVyh qJSmVhnpGlc95XEaYqIhAdDRoKGLytgMyjh3FhuLKtmuCOZM1BEqcTH5ccvgvPuPf/6jTQsBI10J PTZVGQM2d5V28lRS2bgq1TXr4pH1F69tveCLV17Bsxd891sMKr/44NHZHlOw/3X87sy+/MJ1r73x K8bTdB/jYm4ElSEvt6XE0zGVDXSeji7DIwPLSmsS1rSMsATwQBAS6VdVCmMML2lBNplMUEJad9gk 0NUsQkNA++ndAkNYaCfwo4V32B1QeONZKg98tVbLwIHa0SwiDnlPD2NsUMmpo2NnFbLXOoh2AyRx EkMortiJ4huLq3AhysSiIAkjAoeEXjAfdwu8cA//X/xenqOvEmBoVBYjlKHLyvw+GCctwmiB8elo ITPYaJXB4Pedr8YTR49RTd5zBFzsmuKuMWB8qs+d5uWOHmcv6i17h1M4JQ8Ya568Tfbk/WzbQ8u8 rMpP1YSfMVksy6oXVySzbqVT3WW51LbcvsxxsUvAnFopqFU6hXWucju5S7lNNyTe7v02ecpxwPI2 ec/4vnia/J2zmHuFXlUfjG67+iXhNeMpAbSdoL+NcGoqKUqQlHm16lYyW73Q30E61CvJANlu2e58 yPKY+jHNqOqAekTzY/JHclx3WlOmOiZgJBwTSD/N6dzRQN2IoBS28GUobbPSrlrMefNy61brLuuH Vt5qdf+C7uAoHAMlwlND1UKzd+U55jyd40vcmFJE+KnKFnPnjTa8wbbVttPG2U6XlQ3SwPywiqRV O1UfqjhRJatgJKoR1XGVUvU9g5VH2ylfcZWyOW2g+5E4ZBANkoE7ZcAG2hM1zKWhxddSsl7AEVgw 0U9Nl/4eyMbB2mdb2AYoSyUGTEAisLg3WMHiTtBt4qd7QP2wjcuovp6ugLV07VciTEh/N3MRWMht gMXoBHibNpjXyVV5PXxUVOvE8kIxoyix1108cxevlc40xTNN8UzNzmSDOm8VnXmnZMrr4cPA4Dxb vbu726K0l/YaFLWYmWqxcCBSjHm8j1ev3rbs9iq/9Y0HH//krwcffnViG35SITpX1S6+lUz76TXX rLq+bPvvMH7vEyz85HsNXaF6+WawiRYixG1W3IUSRFWS7nAV01hVMlU8Vcy7doOHb1BilSGOVSxO b4a5/pNspgJqMDPRLwbslVRBqUEraVShsM+OkDFuHMXuvWYl3Z03PiaONR0dF8eLammMmtRHxFfp cYStlJQE+RAysnsQ3Cp748oQPEkVx0wQsZJKIGa2NevGu7KWSSOrh/P3mY1tMFRVTimhD2gCrz96 tLh10S1P3yE9ZH0ows3kZurmOG/nbtcpHuZxqmprYFg5LOxS7VI/Ij5iGqlSi0rAqeUVyxPEozLs 96nuKcf7fcIop5L9Qd8u34s+4jOFwnacaAcXOF0RN5uUKkEjAoOP4ov27QS3d5R8thdXJEaxKOtj cWw2msR7jEYcosy6r7c3y/KGhmLe1FTMQ9Usl22eQHbYgCmLLzf0GcYMxwxKg7PyMKfkhFIwr8iU C8aBdZl/2wjZRz0nBlhsprFxYqCxaQL821Rp/cMcjpbZImFrJGyLeVC0LOTB/xF/BkPpnNANXaMP 5jLgCJa2p1JNxIwm8P+sGSv+jic8ffHEB/HYDOfevV0H+i/vasj67Jl5fn8kKXs+5eZPfGewvDIU is1cSZbNadz+wrUzq+p9ucBVFkv12ndmzKH77y6YbOV+DXb5NDQXdXMPyLeYbe0PRB6q5VCVeDHZ WLFxMUEVyqTyoh0S31S38OINdddG+i6m+7xvtd/m2Jkbmn7rrJ1tdyy8z36f46GFo/whxX77fsfr 2dfbxi4+dvHxi09d7HZJ1oyYK6v1X6x4QjWvtsmNbFxtYJ4bOVvOfqdPbbGUqVWDYWwO0yiRGfRQ mJKjTNdEc1lr1jbtCj8TfjHMhUfxIwe6EoPgcEFTWU/bmncFngm8GOACpXtYDrcEoK3sGJ6H58lQ O0+GqnmVVHTmtZfhslGski0bVHirCgomeIwqp3yoBbeMctWyzjlPk3Liduegkzh/SH6OlCBcC1Aj XNIoBecivKiy0rjgBS4N+s4HaR4t4NKyX0zjDemd6V1pLu2g+jWtoyKRzuWT3GAH7qBj04O0QuGN /WIZK/yWRWQ6aPhAowdB6gj7YzjGeNDuyu6M4YWxvthY7FiMjxloS7h0mrnUUPizbKaAEbtWujh9 sXzxbphzxcX0Vo9Wl73YsPP+VtzKYjmt1ZING219tjcB7EcLf5NNLB6po4aBjfXRNkp+KFseasJN 1WmunSPtHEacyBGOTqXTm2U5PJWjr6eGMi08S8fIXb7s4sP4evDtNHu206/RFJdkB8YHJlhhPDFw Qkz0f8ZOEsXtf/3iCbb4DohUUgoTH1EV0STSzd5058uASNtDY9AS+98MfBggoCcGTo/TLaq0Jvxh GGoGqOCxjYPm/NSGivxU5Ghz29KGWaGcx2t3YEUkXFOdqc5Wc8rmyMJIMlwRWRLu8GDPNJ8HteUW SGgGbpLQBYomD2qvWuBBFyU6JDzT0erBndGlHrxkqbfBDc3d09D86nkSbpuXq5VJi0RjznyjB1+Y WuRBi+OLJDTL3uIpfjNgKtJeSs7uQkyUtgKwzYgD9EsDuJ+pNlmTFIFHc6KZRpxO7TGX4u1TcXS2 BYD66spgsORHsWCQnR0lL35qv0Aduwuf2crOvmagPPcMznMdy47uvrX35YSBUyo4Y+K6+iOPz5xd 6Q+kPX0/u6BnwxXf/OKl29u0ppywPJvIY+u81TOz7fNXzspMfp5KN6z+4f6nMtmHf4cvjH+9+84j skKptrs0CuWcvsGDZZF8mUkSeE6h1vdd1L/qnqU1tQ5HeIZ6lb/aH7yUbNu4+ZGlMwY271o248ub M13hdGj61jlZm40HpY/0AE5/B3+uluws6UZvvUwFV9SYNEwRahwheu5gIXUHjfRQmXDQqBzz8RwG yqSOCNWWfloRCWRz0Soc4HU60hlgzwhUOegzqkYL/95Pa6HwGQtbVU3JGBQ+lY1MKbPnVWHww5o1 oGrN8AnDJwafKMqC4jXmWCwrV4uiJm8lTyNZqRT1BkHrfvopMGXJI2RGq3jk1RrxSKJYcxRcxCPn eIddWTMVyRxL4Y3RLDyUPtIU1TD1q2EqV8PUsqYU7WJVpfiXo74OB1h1gFUHWHUARnOKoQ0U/raf XoDCl8/Sa1VV9XUlrc2Udql8lBpdMIpihIzKFaax8lS9XJHT1PeC3WwMGyOD9cP1/Ej9WP2xei6h xO31vfV9tEqux5LKEfeZRjmjbCqvivui88o1cZ84LxiI+yKjnEFOBnPRZHPWl5uJpWgtYqMEs8pk EjVOR0g9rMEjGmzU9Gl2ad7U8BoKUuEqFAgl/VXtVb1VfVX8YNVwFRmpwnQr31jVsSq+qrfuO1vZ JnkaQJtgFijNp5aSx5saTfliDC0xtTmhzOVRqJRhd8SjcHqwoHIJXqqeS9EyFh6mO7JoLMNUW9yJ w7aQ2Eq6mq34FlfAmHMItaVddiWfES/YcEvzhX1ui0GTlienW+UaDeefma6+Yp413zrZcEGwzGH0 u6wpAzYr7p5YuXnWkkvk700+v1RyeEKhaES8EM+8/9JUduGk59KkPxSyaOqXcBcU/UfqKzZCIoC8 aFE5Ka3PHEIhUARetiFKz9hdH2CxjADbsBCwODg1aBCG5VA4zhhfTb1Atq41WvjZQdparXdMIT4U fr+/JG7Hp8TtnQNM2iQaELEvDGwIbAU1XL4BZLhXiZXMkmV+O32AslxpAWvwHQD1oz3iBz2lGElx PeYoiARgZuII5bEpSdBLTAYCLKXP2d/WVio0NxcLsrOuTtkp03DXbiWhL0VICpQLFjq8z2QPvVOt DgX1TB70hLK9nskDHVlRHhxU8Jn8QM2zRREKBc+RgaKPCX3/4GjT0eKSRUkUnMMh3BvqCw2HdodO hRRSqD1EZJqEqMKsqcmyvL6hmFeli3kwzHI56XRlQUAs88r1cZ8ZxCLqbJZ8gZk6p84yDEPJI1Su EyxmzbAaq/NUB+9tydFMNjbluCt1Or1TH3LIibyDrR7VNmSHHbjdgXsdfY5hx27HKYfCsTe499tM HNiXKKkMgOodL5qpoHnpPtKSMLAh0ZgduGwsNHzu14zO8HVp82iJr+MV06ZVVDROu8lZ3TzZ0pJ0 qwWfyxMz4DLF3fRCY0XFtMnAhLQkD4zsauzEK+6rlJzGUB9bW1g12Yp3KnYC38bxkRLSa2MW5gZZ /JSCp/dTiGaFEoMen2LQd2VLkUOL3K2h1Xrw3yfZLVD4lN0Chd+wW/z0FjW9xY+U8SjlWF1MpgG9 WNzm/pmIUuNHaeROfOdoiTETiSnWTLwK3svBb7qw0okTdK6b6nL6xF4AQDnRnhhOPGl40rs7oZTg ZDDBiVBzLMG5VLGo1Bz1xWY66ZCUnRaXusLpluI6wTaKDbJeREgnwJuNuyzYQoNfjRVFQsuzc1wy Ybe7gMJFvmXhPzpSSEN+/7CEjRLulXZLpyROkliEcLTwD/AZaXxwb0XirQClOtu0VVpHoAtgF4qz Lpv50YLTQH8wt2iAsKkoaSPKo+79jOPGB7rpFvHiVzHy5kQenY2iiR6fwegNe4x+D/YZ3NTOwVMe DCgKcGH+g2XOWcoqfnviHM6JJRobE8Agg6/tvrirOuBym1YEHEnbWf7ZyS5XJBonpS/XfHJiRjBY oxeWhpd+jdz1QCJQ4iFMf8uM1wH21XEvljgo4WImgJOlEguQmYq73FgKNZQHbDQF++BjxiW0ICeK hkJtNOnHJROB7U4MKJnRkGQ2QNJG2Ss5ZSskp2yFJEVT+oAk3bzI9psmRWzy8xGN3RWOsRdRs/05 sBgiKAfcZ65lFkNtHYo4dbriehn3+4NqnZ4xOPf7PRoljcCNJ0qGxERibGzs7MJYSVe/CsgJHEp3 fydKX2hxy4eMeX+emJUihv+vq+/TDGuHdd8wPmz6hvlh/678Po0m78y7lovLTcv968UNpg3+bxD1 J75xPxlU32x4lXvVeJKcNI6b/mJWNZmaHE3+eqkp32oc0FxrVKVIhSiFpUgqX4/rRcEqduKLxA6J D4pL8VLjR+I/RMVc0xz/y+qXNf+tUdjVNtHv9ftnkRlGpdZktOhdOq/RZ/ArF3Od/GJFt9hh6rAo nUav1+dfTPgS9KdqHYyrschpojmYoxt1WHcDSIdG6YzqdPDqkoXDAoOBJF17pOfUcGZYDoV/MyxP JvP1Z20bZtpQm+YoKKEzC3+gcuRO0YiJyWyxiE6/y+dMgrkSLdcQtU9DrZVosDaaas75ameiFNIC 8oQkf5mEieQH+zCNSRnGBEtI8lswHyVGjSg6NHUI2Ufxp/J8h+6nWq1GCbzvdDo02rRuUEdO6fAx 3XEd6dON0bUdu32XAztc/jzOg3mDQqkUSorJEbZlU9GexIPJ4SRJ9tbnR/H1+wLf+QoT7v6BHhBt sDAvFAc+o8XTPWD1NJ6za47KspMOmTpGwDhiYyPbN0e3GG1TlQoIGjhKWqC4ja745UB67Ygg0G+k Dwz006WfAdzD/lA/6mexOBHEpgx8Fn8MvC/4eGVgvJgxT9helryWZqa8sZipixnd6rAH8KW4i660 SkRX3k3Ub6Fbh3IBq1IpCBbm12SmvnOIqQ6yF22runONq4Un5+lUgQi++6Krmj/5ZGV5OuScPtkS cccm/+hMLphMtgatWqNBclkrTFhU3P1l3y9nmnW6Mi+RJJKc9t7kr24IpAyaUAhbLfYMXjt5rLve gUMhk9YeWMTN2DXbbQoWseYCUFpGwBor/tqUlWUHI4NZWWU6JRZKUTqGGpihBtZRY7u0Kv4n5mfo pgwpHTW32KL4aOG3B9g6ueKHAA8q+p1bZAGI0FrOrJAL1PxI1JwJ4hUtkyM0jneO7xC1MFupjC0a FX/dUSjF74qRO6ZIaKeKpo+uCF+sUDR9dDq77Tzzv4mtHVFUeXbYPmY/ZefsLGTWmqW53JCflsX2 vfrVte12LNvb7b32PvuwfTc0FHRxnzCvHMd9ymhwaskcuiQoNQiH9LrSY4pbX3LTssM63K7Dvbo+ 3bBut+6UTqHbazvHeCka8U2NZ80VcJxZFI1ZK+dbKFO8cYMzO3uyqSnpMvgdrpgJmxR3f9G8pN7L rBFO/sbsog3N9IgyzT2DlnK/KOkRezfzObtZJNZuYqQ1dc5PTyF+mhKUko/WyEZK43SCtUpU17VO tWqdakVr5ABt1do8u5m1a2aM0swYpXl+GX3b/Kn75k9pmPlTD4DCv2UnbTtfQx8zP8FuT7DbE3Vs vxWtqGNfc6ij+6TYvrM6D31wHXOFadM6wq6zXfp1JvYME3uGiW5uKT5DSpfWQV8uPkOqYGuko4X3 ZS1tKpHS9S+BR+m6qc2Zqpk1hxpV0uyOTpm2SXXihZ0bOrd2cp1LlLOrHeFKrdBYqSju8UhRndbT A8bVxBj9m1JpZ6yu84olVqdRlSNiguWvMl/hTOhaboTHw9O1gkLo6FwiOKpnmxjHmyS2kColmCuc YHWJumZ21szOmufDOP70bHFptauOBhNodV0xqsAKf2NX6+q65lMtTyvnT0kQFD5nV+fP7+4qCY7p TCpCz9kHhoDYmI82NVFYBu4d0bd1dL2IWgsfo1nwScEnXfj4gMvhdIALX/zrdsuerHCs+y82bhBY vJv63Ak9Hu4G11qK+xyj5Mv95XVxXzUUZG35/Lhv9rxyU9xnB+96fzAR96VHOf3+YHPc1woFeXqw M7qgucPXOVMVr1sg5+MxFRLCs5cspYQJV+o0WkHJK4TZrdVph13TDRaoaAoF0hLuk0YkIo3inGys iycTofp0He6rG6kjdbTOtmBpc2j+fP+C9gVkcMHwAoIWiAvIApDrg2W27ILeru5Rsgy01lbHKF59 OzNLz+xuOU298xPFrPFCap+iJrag28T+FzAVNvXbD+iM3z7luZeHdEZ9OBgJ6QIebDCWG8Lneu7g uNOf76HbXOqKjvv/jfte0iYsti4I9rM4cqZaOMevP8+KzeD21eaqdZklN1rX3t02tz9g02tqL5hs tEwL2DW8O7okd+V8QqwNrZPV8/NaRaByYW1ucZWzum1yWlONi9m6USMuS5BPVxsjFauXX9/W1tlw 4+TGJZIN3Hy7GDS146G+pJybo01MtjHfH/TSRVBXLXsr6yaty2rdoZB7Wie+9IHKszaxDiHun4Bl GXIGy3IMy9LMJq4uftVDZbQFKSgk6VnQG4qrGCiVvpHDEEFlY2G20s5VtkPCNgVQtqkNnTYaiY7Q 5jbkZTd72YO87BHeOIuyxZnxHJ8ykuNFM40VijAXp+imoXfEkYeE0hRK1NXMP6uu0dOfohDhU16M u8nqkDFUI7gqizvGUikWZBPZvrH8+ebxOQgiUggRi8G2s8BxacrGovQsDl7NyqwD1cXnG0Mqpj9V DCtUDDdUNrYRw8aqbCpaZbPlssjLWnpZhZdd9LKBsr0aU4ARp3BCW8Tjuez/NugG9mlDTq7IqXIU AdK59lxvri83nFNU8Vhm5UE4G8kpR3LHcmQkh3uhYizHeVW2uM9YDMDF477QvHJV3GeYF/TGfcFi AK46WtGc9lXP9KBgTYaNOBQMGo0Gjd0WEoZVeESFjao+1S7VmypeRQNw7njGG6rwx9vjvfRbk4Px 4fhInENxMU7ibCEYRD7emy0G4RL/+yCc2eHklHzYydk9WKF0KFxTglz8yn5PP915xGJw/48ROPrV /HMqz5oBGdz26D1t6yWbQVs9Y3KaRc5o+OYF123UGqgolrVWG/1Tkjj+ctuSxhsnNy31O1nszbgQ X7el/5ZJb4/NC7I2ezXueHyOi0oaAdg+wR0COTMiL9GVJM0DhmBxoyMz6Ip+nUi3RutcPJUdepEW ZAut5Fkz3h5WacUwKurG4na2Yuzh7CYLNb1O27nozW7KUy6+jHFcmU5kNpzIDDieWQK0yPM+na64 WYIpI8pcoI3Q1HLsLPOgFT9hO2h7Bb+uPuJ9T600/1GD56hn2ZZab8d3qbcb33MLfrkmx7NNErv8 +FXr6y4i+/Fc1VRvzOwHJBLgAywEVuTxMZq28718Hz/Mj/BK/lP64zZNsm4XuDln9gfQXcI0QJto G4ktbhtpX7Rsj843d4+fn3vRsq4f0n3RiIePvzBGlWBL1/PIxdUgHpVxNSfFk+5zTkE/dJcGRL8a gb3msCFCwp6IJqyMmIxlEvJil4Rtaig5BChZ9KKE3RwkVq1dQk4FJIlzfyUCOI7uCwZeA67DLV2y 6VpyrXKzZrNhs/l627WOaz2qnu6e4rcj1B7RlHfDx0oXbLTFBRsaOiv9mETxh59q7XTVtsxcWngh 6NhNV258c+ubm9du+eni3JUzdt2y4qbLZ3PPPLLtmRu+HHx8x/dv+td1zU2P3Pja5G93/+j0Xb00 VvavyXncYeC1KMqT8hKvxaex3fc1mgqa0WUBujJicSKJi1sYBlsktvleomscUxYbw13pzI5ciYsl zLxB6Tpc/LkfWQsGSDJsqO1WCixKpkYMhREG7gSEBdttnAHueVt0x8RXAVhT5+1zO4RqCl8eoIxY o6E8yTaraTTTGqB3jG8tDCMtUlEHsBjWn2U3M9ckaBVTGqIIOw3QGS3tDe0A268rFpERn9kFdKy0 DShBufomzTTKrXlxrnixuN3E31GJp1U2TWurvLjyCtMVlVerNpk2Vd6melw4qfqXWp+e1pXpzq7P 8vI0nFJxsbjZAoaV845yC5hX0SCKBhZGfWgmMSdiHJ8UazHtCRFon5wOQ021XzOsIb2aQc0zGk7z iURYIM8tSe10C+tgANOtn8XtnopAbwPd3MvcGfpV/tK+XgqHNBJrPxOJ5Qz011Iaiz+TkcoJelU4 G9FF0uGcUCPhlB6SjLpWwtXapPQfP5PB1hOBBblwxjq1daC4LBidMmEytnPWJhRFwKRfFSmZOgS7 IrN3Lhy6pP/Ovu/Nq43V2PNtk5KzLmqxikGfI4yzasNVi1dPX3SJ3JVOhbj8wDubVqy/7e3xb2y1 GqsmT16a8YXD2KatXs2t7E47DFsnv7ch2NB14ZpDP++/0GFGxfU98izwcgwfmNrZVsE4Wem3m6LM hIg6/Ljkcp3rofinrA//lN3gpzzD1rP9zIHyM0PDzzwT1hCLnMPmpGE3B/q/WvsW6DiKK9Gq7vl0 93y65ydNz0jTPZ+ej2Y0I8/o4xGCGRnJP8lIYGNbThTJCIdPIJYti09iR+Llge2TZOUNToJhE3nZ 8NkY4vEIjIxDLPKUhHwMfmcDb8MJ4E0cXvIWB4c4PBKQ/G5VzxixYc8775w30r23uup2VU1N1a1b VbeqolCd7X2x7bGJGBuLm71WFqrUaTISOQ/jkL/RHciahPTRxbkwiS4K727nJ3iGhwi8Jsgprc4O OtIgefwrrc4KGbkRwUwc1DpGURoSS+aipHlqHTN4uaf3F7eDmi1mmaxYZIrifzGYiw14qAErpC5S vf6+cCymdkYDsS4kWBocblXCBu8kmTWWrNg6wLLIDJr7kAkXTdiUVhpwA3JEFEVR8aR6QGWQKoEm P6eeUY3qcOLRy7s2dF1857kdOytbeHeeH3RUzltDSxYIdkIvDOLN01q15K1qx7WXl70+MjXTO3Z3 2+rmSHiTx+lpbHLZVly1mFwZkgWjLexTYgL2sEdffPHqVKy125341OKa3hh0sZEaqveOHL6yjiq0 GN146RzzMtSXZYbmSn2J5Wh9yRVJH8pgurKF6coWFv0+LmYl/rGgSNYLSJhIxF2WhIvLzFxMDBqc SSO+24hvM2KjlsEYN5jlOwN4JIADmurDw75RH+NzWlBhfnAQeqoMUCCDxJSWVBHonU//4rT0C13e Xa4d2aAY4wwNNQFn2sg0LDPr0cjOHiP+jPHzRsaoNZi7AvjGwK4AE9CcFkxy+E7RR2qLKOayPs5O dc2Yk5BYLJetyLV5nc4TK8dBAtL8/GBBmqc7ZSp7FxJ8Sk4xTme6aMmn4pa81z1g3RJ9SDoYMQpm IS4khnOjucmcSczNYrW4F0Tkz2w/s89H5rX/EX4l8svUm4Y3w29Gfp+yOAupwdRnG/ekpvAUM8VO esjJkZN1+xun0jaym1RgeaupTki9EPpJmKtja9zOupp6OeFPHeIPCQ+p94fvj1icSVs8tTbVlxvK 3ZW4K3Wf/fHw0dzv2DfrrAluWQA9xwSwgjP0ILRkGT2XnsW+oqPBG5Cf8wd8ig9LPhVKjgTKz9WQ wJDTGQnbLAYxRokxgH+M0pmGZQiRQvV9QZa9xOTeXZMhBcv83ImxkxiOvE3sglh30TIq4mFxVDwg suIsbi3KMZ+cVjjMpaZjeDg2GpuMsWqsKcbEnsUqymL1WE+1cZA9m1SFXSA2i5eCeHAgn4Hev3wJ g5MevHqRHldIlNtzSzZzgu4ggDYdsVncNpulurVzQN/bObjzI7s7walXoqfSKm9rRskBfd98PKGo ksNkVhwwwDUluDpEjphD5rixDuuiXz/UiuwTfN/8rvSu4/24YXAAhr5kA+fmojyNp5lpdtryoO2A 54DvgP9A3aHQN8LTjVZQYpLE8oRY1BQtmXAm8qXUQ5GHUsbBAaLaOOKqnOfjch4XhTwD4NfNJ310 1VXIp8ErRYHPW6WAs2BXCSJbp/15SuR8RDdCDevECuS4K5/yuvS4nHpcohOScEISznxKdZJ3LhRF EdjEPCvZIB0bieBC0WmDdGzAA+B1UPiP26o++sH6PiuyYfXyoY61lw/kJKcVOXLVbS6R2NLNqsyB YPTOT67cqCpDX/3Zc+Mbbgt6am3BYN23bujetHXx9cbGhz7fui7nkJxW9ujiC/ffurZxeTyRXjXy 8J5DAcGHV335767Nd3/qQHt+044HakW7F2SY+9IfmQ7D88iPF6r2nvVFJ8iwerrcabHSYbLV48JG F3W6aEfmqtq2uKproC5SFnQE77JwKbHGbSCGnuSCoMLphTOnM+fnK33Ya9V9Ux/KJ7lWt/ai2LPE 7SfrU8ThqzpkMhdIp5FHLdgi+rHnFjde48Y0uSJURUjb4sdGqsIZ6ZDXSHtBo0sf5JtoTmn/56qu xbhc9XVLhrzUaruwcGZwcE46Lc0PVleg4Wf1n0A2yECnNT+EhximUH/IcUg+5TlVMyv/TjZP1+P9 Ptxn7bMNWYdsf/bCeNHjjXnZGo9X9rGYILf/MGY9TZXcsk0Mg03WFpLpmpc8b3je9rCebW7/z5GF rNCkVOg805n6Uj1TjzA2GIwRd78LT7owckmukmvOdcZ11mVyDdcd2V9V4Bb0bY2DFweJGdtFsrtx 4Zy+5gJB5zB0nwjAmddP9KFn+lILkpwn7KAnv7bl6P6CKLH0bKVnLK595ZVcPHiVIxae7Epvbvj7 trHG2oTh+cV/Wbnw3YGrEvEbRnJDI8zNwZpbVke3Va58gjHoAnsQaUxTpV7VxOhcD1dZjLCo8crc bUUjUgOVkcA5fQVd9VFGn5POEzurFc5ZHTOA4yI183BGqkMEu1czWVS711SfslvMxJL6aTJE4ASU eS1JrH1BbShI59+qLqRnqnthlmhSm8y6iTnLCRbV4rVHtFqIVY/Sgjm6WiHoqxV0/UL10bULH1Wy fAId/zo5LqrSuqea9PnbqJOstxAWZ9Wqgzho7XM6Y9GlM7SAJDovdJoaKhM1HKohVcVAI6T7B1pw jEy4qDHSQ5RihmZLm9KurlZWq0Yf5+ojI4RgX0CLhbkY7jQHuC7VotVzs7i76BKQpkGnRL6PXbAI FkuQbnCxoxLGIh7F0/glbMDUpMkp+yJOZ7/rgIuZBFRysaTaqZWKB9Uu+oOJj2pq0BmRfS9SR/VI N/JHJ78u62pkwd5fJzrqRF8dkhx+qb4O0QkVstUFDyarSyb6TpZqTQTNzdwSrNRPB1mLY0fEYI0S sy/+ofGO3d3rdqTq2lbjzoFC8vae/Bb24MLL03T/yg8mVwx8eRIf6sz6sbbw0GR/ay9jvqaN0cja CtTR81BHVeZ5vY4e53nkc5roCbMOABWAYX99DJEF7/NvvVXIQJ+Q+XDuY5lX4P0cz4eC8J7FTSfp 3C6To4Eq3E4TQ32ghavUoZJ4Tic//NdNQzOvnZboPqgi71wvbPZ+QmZlsuPO0hIi/dBWT4tbdvvC fEgIOlRnxKvKqq+dzwvtzry3RW73reXW8F1Ct7dbXuO7hfsH7hD/Td+D/unQP6PHuUf4h+WHfY/7 v889zR8XjnufkZ/1nfTPhV72viu8633f1zjN4xC1CRpupjS5TKeBhE5XrdJpLKbTcFinDgelxaJc 1yyGdqOdeCczatyt3mO81zEV4tu5ZqHZm/f/yDQX/FefeZ+w37tXZtucq72My+sOuJBfDSCn4AhA K7ivmOJ9suqV5SZecPO84Pf5IjwHLnqpioEDpczlBMUJmXyyxTuLoYMaErAkRIRp4bjwC8Eo7OH9 pBJLRVPmMHeCexFa7x5eHveR7ewq4iG/orOZrxgN0/XebAshz1hbED8HA6ZZfOq4FMKTIb00gIvQ 46KrOUhEqywlkzt2XqQn0foWvG/KUOe9F33nCd3pPa8PTmhdJ/J1r65Q7TVWFrI/7qBbskw4SIz3 KntISNVPYmJu+7Sg1tgKILx+9wxQPmIhZlRnQU8RiBmp4MpzKigqAJWVaqzvBiFHSBFVwuWiQ2Z6 XJcJxkA4jImBfsyBj9bFEp6XX6nlLKFmnGx2h+sWTyYWT9TEFUeWPahF1XDToomxLa+386JF0wyO wMoP/sAaWzMSz1GZbrt0zvgUtJcUe7rSXqLBgMPOpMh0jB3xUS9niGuKSTSRil4oZDK1eWnhDHzm lrSaEygKPWgXtWero8MKir3UJJzTsTfKG1CcRn53CqfQuIY1y3gcxy167KlUYzCYbqzMIJK0CoMF YrxHE9ONJ2i5+o856fENdYWWmhgMMh1aTE0PpW/hR9O/134ff097L24lDGVXC+V7wa80B9PpxI2t 9bKs+MNS2iBE66OpaD56fe1jtY95H4tyFq0t0hbrQ714nXkNtyqyMrYuvi6xzzwpTTq+ou2L70tM ph+UDhJm7aR0QjsRP5V+QXsh/kvtl/EzaQUZDWaTx1DLa+YYHzclWmqvlq529BuvM2/0XpfYb5mS 9nn3y/vD+7R90cl07V7+vtq9UdbGD+A7pTsdBmgV8HtqmoDN0C6kWkdAUsPBgIoSqQASBXtAVORA AIb2980QQ6/ZS3uKRa8WUTkzx5sjibg7kYhDfdBiTRzv5jgeNBTZExE0tyBo4UikySu7vV45EQ3L MGCHFijA73ASvwXNKIDfmlGw6CBPErKDfgL9oCTBIF5FDPHEKAUs0Ey9J/GtSEMcfrQoxouQ2Ugk blE/ELcJMK469tQc2pYIk30OnqI/0y/jwzJ+Tn5JfgPk3lcjGWjg/mdUUcMS/OgV+37tJJZQFHmg jVuLQmYoiovRSXKGHn7rKX5PLMM9Cw2dA5VKUFEcT8YvkDMVofeHV+OHzXTqqz+BJ8mpilJCTRQT pcRc4kzCnBhuvKw5nSerf7Lv/MI5GPjsqLRu8PKBBwR7z/lAnSJQPdrap9uxEDWrepiE7j6vj7XI bQpUDhDLFq4qELilPsn//AzspdgscR0cvbGCrDjoCw5JEBhEWkQlt7VABidki4mLSIr6fO0S4ibk Qrk2rxHioU/HPHm8dCcZsWuhooNY38d0wVEVJZVnHGZ1SWLDk9ARz/+w2Rur6cBPrQ64uTPPu2N5 HNyUWHwx8dvFP2uLr9Yv7wCJYgjUKamFP+In93bU2llNY2ulsNuz8A5+v1V1BRhNs93ywb8zaxae YZk1OZuuN/oRYv8nyJjl7DsVvdEaFbzNUUMjgsgyIGmeanRJzHJikoYaAw5d1GQyRM7MUaTP/5Lu dK+zW8BTtin7lGNvdG/zK5ZXal+NvZrjxXRU0CwR605h3PJm1lzXnha3tBrSBWNBKjiWRwvxfHNT +xpLn9TnWBlYE+2N9zQX2zfKG7X+9nHzhGVCmnBM1EzUfs08LU07HvOejAbsRlESHWJKkRSHkkoI idpMuyC1X89vae1vr9qORSDfdy/Hy8kXuSODM+los1cwoDT5DoF0fX0+nW7PV0VaJlMo0EPriUyb 0zH5Tg9HoXXW1tTEmptbBIvVmgMVxGyWo80tzbkWzTlVk3FgRwuopjXW+j1yfwAHMtr28ESYCU+F cVjW0ul8rvGdRCKW64fS3tOCW4xGsyabzZEWzd3SollrYrGmnNWdy1nht/fy1tpcTJMtyzNRr8Ba m80tYh2uU+CXyKTJzwCduMNBeua0oRE3NgYC9YIV1Mynt9fgmrQ2i+0zqoxlIlmtUktRLsln5Quy gXiQHlk+ybSiHDLjm8ot6RhIhBmUw7mTzPMoj9qZdTPB03TrziA5nXkhOZjccf5i9WD5wWqPS7ZZ S1TnHCTKFB3eVA6sIleZ0KZGHNjrzO/JeN+Szg2SMj5HC9qZH8wMgo9EH6Xdb4HLzEkd9o69dqlj z/w8IfPcvBkIB77U5myQHlJRNTWzQKsSiEXZe8/w+Voy12AhNwIB9ZChKl/nKNiKfqngJb7wQGjR VWsvGMn6k9kLqJW4yCHxzwBNxEUS24XjYl5TRdLp/2tZJJtDzwLJkhNfbBBgoz5kjiKqEnCAn4O8 B2ojVRTKTp04dLXBb8tLUAAOgNqiMy9JYt4BkCp68i5dLtToxEmvLiAzIBeKLk++lfPk403ufALA wdUQWzmIrCafKDoAPPksAUi5lqQO4FxiRPe3H/S3MyFLAqggqk6HUIu7qg5jdl22sAPdJvaRCwDa iK7jx0cTwbClprNndSiKW5dFll2/59yG1fnF/kbZVbzv/q7GxsWXI/7olrnvrr32ShBNdbXerBS6 +eYRn6ceBJM3tPOxxdm7l7GRiNteWzs4P/8JhzfGRCJGd/2dlz64rU23X1hcyV4E2ZS9vNIFOmqy gUV3xXCsHsYNdL7XTUSTgzodxMlQJ0OcWerMzlaHFMnzybfgr5A5PVgVWhVZEeCTqN7tYD6XxVnk BAER/hxJQ3S7cwg15y4rPq8NzsPokEoHfaq+JPVs2Pwc8l96D8mXLiAfCHtBqpjsHOHJPi578msJ xtWcrrmx9YvGe00MzxudnMz5+KTbF+UjzogvmlyOW50t/lXOm/mbhVvkT/tG/Den7uLuFu6W7/Tt 8t+V2i/slx9AD/Df8H09eRKdaf6tKQx6STKZamgQMNXXZaLkp7IVJT/KqbLP19QguIEhlUxS9T7Z AK80+HiDwKWAyqBtcOGKoh+ju6Aht7FMOF8vNtfW+mSiMfinBPyGcIEsbY0KbwussKfA9/FDPMvv 4Yjxen3yFZEYoU+rjDo1lMKZVCHFpORc8z8TMx9i4jO4c925wR3nFi4Okr3dCxXTnnUL55K6QLl8 4h23pPcm5zkSa9T/eweNdxDhkPzPFHKqkZuWGKCT8WwbjtKO14qPeBobg2+cdpi5UBI3aHEvLy9+ qfXotVf0tjUF83EhsCrSufiMGJSl2hzU4lh9rHsxi/+aiDt5iw1Udm/QXvjgs/fu60o15GrEqwam mRklHbZKVlJ/L/0B/9DgYOyIRfUnYFwbKwoI+Yz4JsOq1aAD/fYa6V2UWUdMwIItQYPj/dcMYfzD NfTiPNZ6+h75PX5I7PgzVBV6oen315dMVUp2TJiajFPwyFfuBiftBZmDi91ok4Te37V4ndR+OeTD O2TzuI7cDc52oXuBfpb5DloP9O8NCPUC7Aa4lsmjkvHHSDJuRCGAteAOG36DGgxjKAju1fBODtwk LAx0EsKuYr+C1hjQpfeBroQ4uoD2Al8fuK8EsEGcHUz+0gi4HeC+0gR9BritAN3w3l+A18bWoxsh zA1+DOGDtGxA/TrfpT9AtjvgDyolbsU34NeZ37FPsBeNy00/Ne/jVvApwWBhLAetX7X9WhyWrE7B lXJ/0vOj2uPe9/xddb8KLFe2BA+EvhA5F30g9mKiuaE39d3GVzOppvnsuy1drd9unaOllUHvQBrb kZHejpNBGxEy/lR0wS/I0BJ2VsrUhLwIbRjYsvr63uSGW27fNnbNtjuv23771s/2r1+3gZ7bQT+X gujGj73APEPv6jIiEZFb2H2ghdWhenIWGyjdMRRHCdSAUtBTt6BW1Ia6UDdaiVah1WgNWot6UC9a h65BfagfXYuuQ+vR9ZDPTWgzGkBb0CfQJ1H5WbTh0hz7bzPd3dniLNBkmtJyPJE9QQLKvrrsc+y/ MU9AYgp4vFGu8dOQ18srVlQcrct1x0xDY/aNToF9Hb0NwLCvs29ABulbM/F09kKnDTww+wUkghxR 0GH2NVQCYFCRfXUmEs1On2J/DuE/ZX8CZUFe+0nZ5shChD9mn4HyVNjj7NOVkKdn7I4s6hyD6oPR HOAzAGcBLgAY0Hb2MTQBMAVwFMCARMAKQAagj/iwR9gjkM9H4H0RcAZgO8AUgAFtYL8D/p8hmH2c vRWF4N0vQ/fiAfol9n5Kvw3UB/Rh8A8A/Ud4JnS68vwQUBL+YMX/EDzXAH2gQr8B/n6gX6fX0ivs 1yrPd7Dj9L1dFXqYHSsHFKkzAOEqQBMAC66D4DoIRXeQnJ4CGLNfZG+jKR0DmgV6u06huPaUg2H6 G+2ZqZWzh6FI90DR74GS2wMlt4dYzrC7qzy7dZ5Gdjfw7Aae3cCzG0qliR2D9MZIlQUsAagALJT7 GJT7GL3kbwz4x4Cf+P9XwAcADpMn9k4oxwTkaj97azmuQCW7aSZfzBZOsp+Goi6yn56R67NTHz7x AqmIQO0VKhLebTR02wxvJb7bZnz1OgWuz3Ta2RH0eQAGuQFHAJoBugAM7Eg5klGeZa9Bt3OoaFcm mAl2wjBhNDR1YecpNov6QXQqyMk2og5gSChDHbhtmB/lJ3lW4lW+iS/y/bxxOzvBTrGswmbYAtvH DrFGopOZ23OkT1xlas8dsBy2lCxzljMWY8k0ZzpjOmu6YDLqh2r1m4ZNo6ZJ0wHTYRNPTq9ghi2j lkkLK1lUS5OlaOm3GBUzPtx5L3sDER6AJYBRgAMABijjIfBX2U8BDMGvMQRF8SnwR4ARPEkAZ8B9 FqgRnkTgE4FPBF8RfEV6U6dIQ/oBhgFGK6GmyyHVdwj/BRICEINQO/jaoWzPAr5AXABr4ckGTzZ4 sgHXGeYDyKEEWAXoB2Cp31kAqDWAq2FNlfBhABMNv0B5qmFF8i7zQXFrbC6BSwl8OIEPJHCxo9CZ LYYAOZ3OofCQNhQfesSwPbxd2x7f/oihL9yn9cX7HjEUwgWtEC88YsiEM1omnnnEoIQVTYkrjxim eo/2nup9qdcw1Lu9d6KXbSMDnnKyKUtpSCP06bLsy7aJnVcwR+HrDAGeBngDgEUK4AxAAWA7gIE5 ClhhngTfJ8H3SdQHMARghDeeJOIFsFIJI/7TNIy4SDjzkXAWvvgT5fZcX+daELlDANMALMT9BIQ/ Qbl111HqXwJ8lvr3VfgPU38FcPUdFgTcFirmtkDz24IKAEMAowBG9BK7Cb0BADEDVgBGAY4CGNgt 8LeJ3cQ8CX9PME+wqaJtmUdBNTXQJzkdnNQpMVaoAzb8OMUPULyf4gLFkaJ9re3dtbbvr7Xdt9YW AwcTR50QcJDiYNHSaXuq09bXaUt02iC2WhQEFcBDsYlg/O8UX0NxqugO2v4StP0paPtj0PbNoG1H 0HZlkLxXB23XxrgpthCMv07xWoqjRYti+5Fi26TY2hRbpw1/C0PqaAXFAYr9BON3nhK7RMSfxO9A R2pjcLkjocwyiBJ8qdzRCWSx3LEKyEK541tA/lruuF/5Hv4Lpl0afrccOad0evBFvMZAnv9UoX/E a9ARoBeA3gT0UdSBNaDfLnfcQ/j/Cd5/EJ4fRiGO8P8j6qfvTeM11P+blff+oZy6AVJ9qJy6G1J9 EKVoqt8op86B7/3l1H4gXy2nbgMyVdZIBm8tdzQonQ5yfCtDeEeQxpCc9FZSXA0x3wZ0lf5ydzlF 3uoiCcziq8vhZUBiJJffw2HUT5NTymH6JetRmEZRh8I0036kUWrHIs28DYUo5crheyAW01PaOeV/ d5wkXxz9GYvlbym/+R58v43w+Gu8pnxE+e8nSHGVlZdSs1g7rrwYPqn8MDKLN5aVudQsBwGnUrMM flo5BoVcAl4GH1eOpm5SngzT0EfCEAo/9XRHo/JQeItySIPnsnJP6nskG+h2+MYbIXggdZXS23FE WanNYggudpC7FQWlPbxTyYP38lm8ZuaIsiwyS7LSBHEcOa40QIrRMM3K9W3PMi3IjMeLKfMu8w3m jeZrzVeYc+ZGs2quN9eZ3ZyTkzg7Z+UEjuNMnIFjOMS56X2yRC90myRCTAaCDdQtMXRZVlfMGRik Qdspudgepmf9Clxy9qCeDStKbcmeWfOl60rLkz0lrv8Tm49h/HcD8FRi9s1itGEzVFDida+fXLp+ AmGcufcrfkJ33/uVgQHcU5obQT03qKV318P3EK7dUjKGV3hRzR0Fb8F5lSO/sutj0HAFL5kX8H5k lsBbX/p6z/rNpe/UD5SyxHGpfqCntIpc136C2cFs7+46wYwSMrD5BP4cs6P7OuKPP9c1cJkNhZhR YAOtf1Rnm0EhwoZCeIay9VI2qKah7q5joZDO9AO8hjBB9fkBZbpJjysCSUBc/YQAGxNAERpXhAkQ NqgPemTi0sisCIs0MtGKaGR1hOmYpgFLSiMsx9o0YDimtdHgIx8GhzU9OwNIo+loeICmg/GHPHGd B2pBhYfhgCf5//OzbcX/AzOe2fqrG0e6t4W7h8Pd2wCGS1+642ZvafIGVT12469IgFpio8M3jNxM 6NZtpV+Ft3WVbgx3qce2jnxM8AgJ3hruOoZGujdsPjZS3NZV3lrc2h3e2jUw8+jE1T0fSWv/5bSu nviYyCZIZFeTtB7t+ZjgHhL8KEmrh6TVQ9J6tPgoTavnuhW4p3/zMQ6tIOZIlM4wFgHaw7A/OLCi Rhq9ijaOK4LeL/ifNSDotizJgZI1vKJkAyBBjZ2NnSQIWicJsoO3WAnyfuGKoP9Z/HglSAJvR3gF SiJv9y1dl//HxsZ2ERgfTwLeNe6lfrug0QbX95RWkkvcO0od3aXicNcAnWMbr3yu3lyUTnW81MFs 75jomOqY7jjaYRwfHwBv56nQSyFmKLQ9NBGaCk2HjoZMJOCTm48XO6ZDb4fYcahNeBd8urtomuNA 4Z887hofIx8ECYwB6Mklx5NXb+4MoRHQdjFo5o3IBRAGyAGsBzCi/wb4XwB+A/AnAAP6IuD7Af4J YIb4sI1sY7f3li6S4gC9iMPLZmeaWrLLZ4Fu/bRO12/Rafc1Ou3ozHqBlgs5oVMExRujZwH/FOBV gP8F8FcAI5tlszTy8cpayBgaS2LIPpmP3EXQWHIXPXkPk+LeNZZMojF9thLDL0D3eP6HGU48No6g KOAHAQJM1HeMvDZO6IeM/wfDPDLNCmVuZHN0cmVhbQplbmRvYmoKODAgMCBvYmoKMjU5MTkKZW5k b2JqCjgxIDAgb2JqCjw8IC9UeXBlIC9Gb250RGVzY3JpcHRvciAvQXNjZW50IDg5MSAvQ2FwSGVp Z2h0IDAgL0Rlc2NlbnQgLTIxNiAvRmxhZ3MKMzIgL0ZvbnRCQm94IFsgLTU2OSAtMzA3IDIwMjkg MTAwNyBdIC9Gb250TmFtZSAvVFlaSFVMK1RpbWVzTmV3Um9tYW5QU01UCi9JdGFsaWNBbmdsZSAw IC9TdGVtViAwIC9MZWFkaW5nIDQyIC9NYXhXaWR0aCAyMDAwIC9Gb250RmlsZTIgNzkgMCBSCj4+ CmVuZG9iago4MiAwIG9iagpbIDI1MCA3NzggNDA4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3 OCA3NzggNzc4IDI1MCAzMzMgMjUwIDc3OCA3NzgKNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNTAw IDUwMCA1MDAgMjc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDcyMgo2NjcgNjY3IDcyMiA2MTEg NTU2IDcyMiA3NzggMzMzIDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDU1NiA3NzggNjY3CjU1NiA2 MTEgNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNDQ0IDUw MCA0NDQKNTAwIDQ0NCAzMzMgNTAwIDUwMCAyNzggMjc4IDUwMCAyNzggNzc4IDUwMCA1MDAgNTAw IDc3OCAzMzMgMzg5IDI3OAo1MDAgNTAwIDcyMiA1MDAgNTAwIDc3OCA3NzggNzc4IDc3OCA3Nzgg Nzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4Cjc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3 NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3NzgKNzc4IDc3OCA3NzggNzc4IDc3 OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OAo3NzggNzc4 IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3Nzgg Nzc4Cjc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNzc4IDc3OCA3 NzggNzc4IDc3OCA3NzgKNzc4IDc3OCA3NzggNzc4IDc3OCA3NzggNTAwIF0KZW5kb2JqCjE0IDAg b2JqCjw8IC9UeXBlIC9Gb250IC9TdWJ0eXBlIC9UcnVlVHlwZSAvQmFzZUZvbnQgL1RZWkhVTCtU aW1lc05ld1JvbWFuUFNNVAovRm9udERlc2NyaXB0b3IgODEgMCBSIC9XaWR0aHMgODIgMCBSIC9G aXJzdENoYXIgMzIgL0xhc3RDaGFyIDIwOCAvRW5jb2RpbmcKL01hY1JvbWFuRW5jb2RpbmcgPj4K ZW5kb2JqCjgzIDAgb2JqCjw8IC9MZW5ndGggODQgMCBSIC9MZW5ndGgxIDEwMjQ4IC9GaWx0ZXIg L0ZsYXRlRGVjb2RlID4+CnN0cmVhbQp42qV6CXgUVbbwubeqq7p6rV7S3VmrOp00IZ2QmIQsEEiH LCwhQFhiB4gkhFVFkhAQUDaVAQNK1HF9KiiIjAt0EiKduIDzdFx5+pz5ZsaZcZl5cRkVZRx0nMGk /3OrQwjz5vu/+b5XN3c5955b9571nlsdIABggp3AQbBlXXMr9JFp2PM25jdbNnWok7/1vgVAHgDQ zVnVunrduZS3GwCEzxGuWn39llW7Tk4cxBfkAjh+WLOyecUXV/ddBFAzcH7hGuwwWfk7EG5COG3N uo7Ns3sEfJd6G8Kzr1/f0hw4k3sB4XcRzljXvLmVfqE3IPwDwuoNzetWHjsz9AsArwfhca3rN3SQ fdxEhIMIz2htX9l6S1goRngv7mkb9hFg9DCKBHiJvQMWj/T83x/6TzD3b8/kQYf7ARBBj6UEhpF+ I+7TrLUsYAUZa5sG2ZG7AxCv5aOQwPsB6Y9+hvlzVg+vjX7NxobXR/9E/4T4fSM59rwAp2E/9MJR TN0gEx5WwBbYh+ll+AI64XG4i5yEDbAVjmD7efIibUUu7QQ3tMJ/Qi7hou/CM7CNmHHPdngDzkI9 3BU9QBy433iogHbo517nfhP9mlSTG5AriVAJ8+E57mv4LeHpFJ1HtyGajRRL8As4S2fjvm0QB0Uw E+bAUtzTk7jXV+H3JENXEf0IvBCEBbjyFrgTDsOb5ABdSTfSI9zrukXRh6K4Cr5JD36ohrWItQFu hIeQjm+IgTjIy+QTzsM/PPzt8N+jR5iGQAGUQxVsRGpegbfgffgEfiCLyCoaoAu5Vl7Hr466oidx z8mQB7Mw1cIiaIKbYQdy7BHopoe5/cOvDP8NdYbDlI27LoJJSP9i5NVZ+B2xkXiSTsaRGWQBWUsO kYtUpCV0Fz1C/8bpuAxMhdxhro/7gPuI+ws/g9/MfyoYoxnRmuia6Obowejp6B+RpwpkwGx851K4 BpqRqhthF9wKe1FaD2N6BA7CE/AcRKAfBuCX8BH8Eb6FvxELySOTSSlZRa4nm8lx0kdOkXfIe7SR NtPH6VnOxy3GtY/wwFfy8/gN/HvDMFw8vH+4e/i/opZoT/S16FfRIeSmgjxPR45mQwhW4sq74S54 EFd8Gk5AGNMA/B7+AH9GzkmYZOIkbpJGxpNskkMKyTxSRxaT1aSDbCG3kDtJF3mQPEzCpBd38xJ5 lfyOfE7Ok2+RM8hmaqRWqtBUmkWz6QQ6h66me2gXfYb20RcwvUt/RX9Lf08/oX+hf+dsnBNTKufn ZnCzuKXcem4zt4Xbzj2N/HyL+5jnUX5WPoPP4m/jn+BP8O/wX/J/1xl1d+ru0T2g+0T3iQCCLEwR 5glrhHuFiPC+yIl14ipxu7hDvEV8Tg96n/4Z6EHr6EZKxxr2UngMfklegg/JUc5Jnybz6JPkPmLh PHAd9x/kv3U1cDstpWFSS13cX8kmsgniuJ+RC3ABnqM8/S0J8E+SQ/ACWtJ+eh3dzFvJ1fzP+CHS wb/Hc3QQjtKv2TqCk38SV9uEbmodmYqt1bAOHqVOeIseQSm0wc/hUUGiXSj3A+CnM2AimclkQ7+B L9E6bKQMrkU7GSKHdR30MbKV+5yaoJ4M0Y/IZF0HrBJk2EV66RzuLTKIlvcC6ksNWUNLyHIYgk/J 4+RTughq6a1wmF+t+xX5gATIHN0a1D/gP+Zmcquogz7/v3zXCTiJlnAWZnOvw1JyN1r/WRqAmXQ9 PMK9SP4MJ8nN/GpuDe5yM+XJrWgLz0AvN4M3wjQ4yZ2El8gx7tckACf4zeQGck+0aqgRvhOO8se5 bl0hnxR9c/gP5AnybnSA/gWKom9yi4ZXk4f5eLTLm9F625FDRnga5z+MHuMo6LGVjvZ4J+prHPo2 Ca28Gj3XbLiGfIsWcytyqZBkwByaCtfRclEVnOh2xwEEyxcGy6ZOKZ08qaS4aGJBft5VuTkTsrMC meMzxvnT03ypXlVJSU5KTIj3uF1xTofdJlstZpPRIOlFQYdSJJBV5atuUsP+pjDv982Ykc1gXzN2 NI/paAqr2FV9JU5YbdLQ1Csxg4i56p8wgzHM4CgmkdVSKM3OUqt8avhspU+NkMV1IWzfUelrUMPn tHat1ub9GmBGwOvFGWqVZ02lGiZNalW4etOazqqmSnxft9FQ4atYacjOgm6DEZtGbIXdvtZu4p5K tAZ1V03qpqA3467CCb7KqnC8r5JtIcylVzWvCM+rC1VVJnq9DdlZYVLR4lseBt+0sDWgoUCFtkxY qAiL2jLqWkYO7FO7s8507o/IsLwpYFrhW9G8NBTmmhvYGrYArlsZdm8d9FwG8eX2itCesaOJXGeV Z63KwM7OPWr4UF1o7KiXlQ0N+A6cS9Ormzqrcen9jIueHNwI2z4jJUbUSl8V62m6Vg1Lvmm+NZ3X NqFAEjrDMH+LtychIdgf/RgSqtTOhSGfN1yW6GtorkzqdkLn/C298UE1/sqR7Kxu2RbjZrfFOtIw mcc2Vo6OaS0NnbVq5o+yk7Ad+WaiGoTVFhV3EvIhIcWsWFkMnS3FiIZPA8FZ4RUohrVhqaKpU57E +tn8sC5d9qmd3wGK3Xfuqyt7mkd6hHT5O2BNphyjCobjl9rhQCCcmcn0QqxAQeIep2rwxOysTRFa 6GuVVayQfTAvhNMaJuUgz71eJtV9kSAsRyC8sy4Ug1VYntgDwZxAQ5g2sZEzl0biFrGRnZdGRqc3 +VB9T2qBW1xY7x/9s8ouR9WaSWHi+v8Mr4yN1yzw1dQtDqlVnU0jvK1ZeAUUGy8eHRtphR0VIS6R jrRoIqeNoiYuHUVmQMgU5tPxT9A0eUVE1KMqaj1ErQ7LTTNiZYPB6/03J0Wi59ksrbo8bWSb4UmB K+HJV8BXbM/UyeGGeT+tWbi4s9NwxVg1+p3OzmqfWt3Z1Nkcie5c7lNlX2c/fYI+0dla1XRJopHo wL7EcPX+BiRiDZmE2kphWrZ2WDK26zDhqSDCrG5KnicTMFIUaVEP6PgImXCSA4PIGn0E4vWCjo1T 4EhFr7TkJU9A/r50qHSOfKG0dqgUyrAt/4jFVblem9eWjgXBcPlHlTvzY1AHF0Hlz7DIewmu2YMx rxUj+ruDs9yq0VaQxApwqs5cZ9iJZ65VpRIxJikkMUlRc0iOKk7XEVWWU4DgIUAU1UtIBqWOEtmb YQV9Qqa+LlWOEDFos0IOrlHtbSWEyFBd5wnkBEobCW6ysfbcHPn72nO40cAQDJYFSuUhKNXtmRDY Jr9yVS40ksbGtkAgn3jz3Ck0zkkFQRR86fl5RYUsFU4s8I/z+7xL8CgeOLhw81P11z374q7lB4Z/ 8/kjW2cXTp80e8nNTdPvGx7SDbiVg1//tHv4gw+2pbiPJtt92bOWXzzUc1pxa7ea+XhG5yH9OtgY 9FCOG6GJZhDKc94MDJF1P0eWjw8mipRKHHcTSIRIMbqgWsAwDtzV/WQ8eAKBQCkJbJ0jn0ef2Ijt nMDWWvkvCLA2I3YwRmwpErtHN0JoPkESxfkk75VhDNV8uoF/fHWUyWVx9HO+kJ+KkfdEeDlYtySb pBvSjT5TetYkMosIOfoS/dXe1V6+ICvTyOdk+M2cFdJTfBkBzmE25CVkBAJZBrPTYDC70hQ3cc93 KAmi35CncEZ3yOoirgj5z2BKjir4C61qCoRkX6uP+qIpQZu9AFLklPUpXMoLdDMG6X4sPYE58qeN gdrvG8/JKLYLtUPYgrJzZWXnhhoH91gmBCxIDNjsJSUsEyxsdncJ/l2VSxqhsTFdEHyp/okFhSi6 tCIUHsouVRDHFRbm57GoQOSEOKfb53eglC0obVd+XmEhJ19zvOWek3U/aZ5CFs2Km1C2pf0u76ni v/a/uiEUPznJdco6xX/1qkdvmba2efHRptvqap7d03D7ArvJkjzrqrK0vJWN8qPHrqluXdQ6/MP2 uXnXFJBPrbJkCVxTMnv5sqcYjyuRxzNQ9g7wwY/BVQmqxVaQwgreOzd9Y1yn7We2fpsw3paTXpY+ Pa4+blWcsNVLOLszLtWBm7RzSWmcoDgo9RFwEgKoNMClKYogOjLA4FGsRkm1lyURSMpJKkuam/RN ki4pKUJOB00gobmgIoHkcERIYTBeygWSA2UwF5ahlUJ9mqZY9Sv/SavaUaX+yDRK1tRLPsPaW2Nj begGe1uTSFGgsYFo0gmgGxgavKxtTECECWaP5RX2jJgZs7J8jokCJWHhUFDjRIeoycCOZsYEJVSS vL519ftn3/vavE3bb5uy9lB25jpyS/Oyg6t2LVt+pGi8bmDowtzyD391xxcHl+Wsb3+DnEzde+du knDjT356/yMb0c42IK9dqM+JsD9oKOHWOlcn3i/wnkj04+Ai2VZQYthno0sT18o3S1vkB/U6wely jpcqSIiG9II1zbLASNJy8ULZxT4O8HbFKMYrGAmHVJJLKDlvcamiP8kaAotsoZaa5OIaprLfa07m Qu05dDcaQ87ZS3IaB+WhQaaWSDiS7x7RTHsaI1/0M3JHFJDz9lVePPTsbzoJeeLp13vIhmvWHVqy ORR6jNzqeO3lj984TuadePmgaWV75/Bnt+zd+xPtS8a1SGc50hkPqXCmH7xIn4T0KSoWLkZtCPVL EjJS93n2xfOe+OkJVIS++Ffj8XaWZbwxYU8CDwwXEhOAsxObNRnSZIwlKOCFcR42eLKAT0zIsnXZ D9mp3c6rikl0Iy/sEXp3MNGp6v2+ZNUadKsFYJWtrdaPrLx1app/aowhgRhHhgZH2cHOiaHGtkFm tMxi3wgw5rS3oVoEkDs8cmeUPU7RK8R4Q7yFMdXg5oX9w9+8uOnV1Y8RuPel/7H8+C1/e0vjyeE0 upDsva7jNFlrv/Wrde/uPk6mH/zq7Tnzlfh7H9lKtiaZ9t51CPWiEY+7Ct0vwQW/CF7rE0kGGW8s ET9yfOTUeYjfXmjneDQqPo6zx7lcNmyDzmQ0cUbJYnO5fKBDZ62bayEWVSJOmsU5kCM8J7jQ/hwd Tq5DpoTaO+LiJJcrBBLfgb47h8U3Eeo86Zbe2o/2tVX+9pLDHkTgksMeRE3JQTYxIyqVWRoqLTvX xkzpkrOzl8hviDq5tFTEzLjWhr4u3+Eryi+aStGxiZrtiPmij2t8+fHkxxVP/oaWqlu9S6dOLHJ6 3kx+82Xuof33t60oT37UM7Glff+Pq0A7kypRxSejX/IQZ/DG6Tzxi0QxKWY8hNP1M0m1/mpuj/6/ bOJqcat+K/qo5/XP2wTeyFuo0+i0UM7todTj8cUOMslk8pllp9ksO9AvMQ9lRt5IEvaHzNIBmciy lGMuM+8wv2PmZfNc8zLzejNvNkfotmB2AropyeNBztnxACf/7KTiJQKyGf2U2V0/lXmqAOPTWG+F fGwcewqO+qqRkdjAVubGYpxm+dwlHmPtYX5LtMTYi087NLaN8VYi53OMKKIgcsxJbVx5fOltd6u3 9u1JnlG5vGdl5jJ0TWeXL9rXXnzf0B301v1pBdNW9742XIy2OgUNNg35jPEWyQkanuR+wX3Gfcfx UiR6Jjg7p7hgrrRTelfiFClHOiidkE5LUUnAUIwnHLISCJdBRdHHEyfraWG8FXSCmMEbkGuieAMv yRrXUCPZCz34wp38uzzlg0ZrAb9Rjy6eH2HcCNMCbQHkBPrxPj5YO6FMmyaV+cv44NR0Deqt8cd6 LeVe7HVmYGH3xYaSc2N1Uk6sdo+gSk6GmjxOg3rivWWBsU/D/5LRiMqz8tzIwRFAZRd1Y/Q80EaK 8kXiyOdIVeBkYLjyw74P+XNnz1508P6Lv2M6XAEg7EQ/mElmBqdU20hWUDIUHMx63nc66x33G77P qPCg+0Hfcdfx1BNZz7uFKku9fpHlavsqy44sDKpS9amWifp8S7VeyGI0zDXLBdz4TEozM5leY/BZ gqaM6ySnpPgU1amyDpUoimq1230Op9PJOpzE4XCmK0K8YjJpui9kKinsaM6KkPeCZqdVsoecMjhk B8VT+LqgWUmWU0KqDIqsUIX1KEDlzBCRL2t9fQAPZlVWnLKDiS6m1bHMolqsaoewiDXHtl4hsuZN 8LnkVDCYwoAXHS1Tdmx5rmxe1vwAU/0AupZ8UVN8979U/7Htig/70lcebGzZHVfX27J7t/tA392O aaV1xxp91/fdJ5cX1P7s2tS1vP9EW/3aa1a0bG+/qm1oIX2pPr2gdPnBJ4aG6NmZSkFw+YnHhw0j /qgEZemGb4Lz0sSJIvXRNH0hrdbX06tNq+gW/WbbU7bT6Ibe1r9hs3AuN+UFjrrdmqyCckmrJqsR VyRjR7tMRjwSFyHDQRulRMgwuc1mMGguRYqQUz2mEF4iTgXROZHLjul5ug3PVkoGetwhEiEDQccY 2XguuyLtijE4NIhVG1aaZ9IceWmgrBTi5UGP5mdGOM48TYzjzK+z8CgWuI5y/Apeo6f58Mi46waa d92TsKfvjriZVfvez1/N+/vXrdi/cfKOoW30seU5E6e9/tdhuxYTrMCYYD7yz4J3rM39YMMoYAFG AYks1BwnkabU1lQq6BLjnClcg3NxXH1KvbI+rkkRKnSkQ97kvClha8pJTpek8CIGPkarCsHsnALw e+NVEGWxVeTEDan+lWMinkBtLNxhAV4bu0ghbQ65KEYI1Y70InaMT6VFMZIsdMWp+7878+VPh7+5 /+a3ruvrWj+pfXlVnHLXDYv2t00k95Cit4+df/vU8KvHrv35Xff9R07TTdNblnQdrHv4nZHziluC 9FnhQvAmifuJdI/+LokXzC7zUf1r/J/5f3CCn2bwxaSQziBbyO1EtFgpZ6RW66VDCoM8Y+x4Eqxo n+g/rRjMBS1ygXbdsqPJQS4L/WSgTdAK78J5lHdM8hzUy1q0/FY/KR29hsnfNmoHTTt6034A9Ifo BkHzhxZ3GQTNLg3qzXZrdU+KveyyP8TpgX95MpVd8oigaYl2KLVdUpLYec+hchzKnv/I/MK5s3KK l71espj3v3/TpnHHUn81fG64nvFrDuoDh/zKgm/7jJlWjA4j0V/3Ys2xADEeG/eaH/Y+nMpt4rbG 32e818QbmaqoONqLtZdhVWJjN7fPc8R41MxXc1uMe41cpinNm+orNvGqycglow/FmifuNNd8B6QR Mj5BcYg6ZbwxWQ3iud9BsljIKJGQyo51wmKioJzNbovn9Sqky+k0/byLccyWNr4AXLKLfowXx5cn 1L8cU7W2QO2FxqHBRmy2n0NDaxuNsVmIbXPjXSN2BYS2Ri2a1EoHah6LINkdMG0ciy01jxaLvPES 6GIpzsnC8nH+RX25u+o3b05LH/5jRkXl6ydf/2++m9+58Zo12Snb3i2sb35tT2TXLnKdcc4N1U3l OZmZN8WPXz9j+8n++01NrfV5ef6EwsUFC26c+8CSJUu0O8jX9G7dMUiAvcHMWdZV1k3WPdYHLA86 npTCSWeSPnegsyMcxFvBbsyymfDc4IzW8zb0UD1yh32ADIODJvY6Q5IpQhN7zB3GF2giKmsiSMgk Y1oWKqssHZA4KUIP9CYW97JPMo2BC4MXkB+s1IJujLfLbFroyLQnXdTonFhQxMJGRxHHgsVYjE2+ TCmfcn0wN2HXgeQDRe/U9aR03+ROzyy956e2iRlVvu107X6i2za8ff9QX6tLTUX6Hkc/vYX3gwEe CdbYx3GqqdoQNM0z3S7ulXaaniBHDaeIUdDpDC5+nKEYdBgG5ut1Tr1ep6M6Pc2P3WIlvZ65YwNG iSHQy3qqj5A4dNHaXes84daTA3jlihoHSK32aYCJf+g7dsEqjQXKzET0Y8I3pLOtEc3wpN5gdxeQ QIOXy3fgGVZE8gmZdzy8xOoqqCMtp4cO8f6hU03vt91Nd2AQgUfGcB2/DelxwgPBBaor1xg0BV17 DDrJZDS7JLch01hsFvR6yWyxiEDiwEH0nFWW80WLUxQtZotBlDmz3mqxGAySoDdwqgO9imwh+Gcx hCQyQO+GOMJ14y1AHsw5hwFr7KOGRoW9xBb7ksE+b+hGI36tR+Zf0ZdqZNmKXNonqSKNJCSI6a2x eGJRalbBpO6eeR4b+f2LQ0uWP9BSNrzqKTneu2QNP37os4MHuasv1obbY/E+uk/TWX+y4ell1tLv 9PF67eehlxaEhcv1cJ2wE29JFKTRX92xFqcOz4EK/ZmLHwzV6J8EL4hX/MLkEBCJluAN6TVYwm+A +ZgXY67E4H0D1tfiWCODyWswRXgKKrSxDbCCPqXhzMHxDVg/zsfeNwnTABFIhN5A/8Rt5TGu1a0S ZgtfiUvFC/qwpEp7pR+1/TmwKIPFoMMdy+inF+ExmClsQ19NtZ3bR6gQIA5gxoLK2YvKA+Xta5uv z57Z0Xz92pbahdp/BUQvYB4H9/3L3/4d2n8KjMO3V0IVVMMMmAk1MBvmQR3MxxXr4WpogKXdC3eU 2+hROIH5NOZvMPOQi+VczMswcxCkR3sO5AcjWC3Tqt45dXk7WT27Nk+DgzNitcEcq6VJsTo3n+Ed 6a3azOAjvXmTYnDmVTE4LT1vR7lMjyC932ilFcsczGWYd2DmcfEjvXHJsWmSk0073JuQmGc9TQ8j xmGcd1jb4uGgAYftc4W5Iv2mvIh8iW87qJU7tHKZVpZpZY5WWkdGv2Cra+VprTyhlTlaWaaVc7Vy vVZq+OQcpq8wfYnpC/JF0A5ZBBQiZxFZIcEsElRIP5GIsadAuStCjMGiAmWCWqHkYc5XpytZWCuY b8qcoWRj9mZWKkUEmHshFPTgZl9f7TZ9MEKeOTW8xzy0xwwY9JX1ZM5WyiWCesaz5QoxP4SZ78ls V17C2aoGAqj06R7lYnaE1Pco/1AietKj/F2JUBJ0KD8og8rflOeV75RZyhuZTyv9iPVQjxJRIjxi HcqM0KeDVmWfMh83N6hsVq5XblC1oeu9WAWNSgtOWpy5WAmpEbbKHFVbZTpeCB56TqnCwcrMCCHP KUHldiU/W5uax6Y+p1yltCsTFG25rNhy42N7y2DVc8o4XCxVW6VKWWSWzFJR1x/ErmNi11Gxa7vY VS52TRa7CsWuiWJXrtiVI3YFxK50sStZdOrt6IktepPeoNfrBT2PXhn0ThYIBJg9OQWZVQLPSl5r y5SVMXPDeFlPYRaEHVwNrVkwjdSEz7RAzXI1/P0CX4QY6haHdb5pJGyvgZqF0zzh4kBNRIzODxcF asLivCWhbkLubMDeMN0bIbAwFCHxrGt3IvsdsR+lGr/7jkRWR3ff0dAArk1lnjL7VFtJdeW/KJpG yjGXUM8VV9JAzbwt/SjlUK+oTBERXIBgFwO7GOhJDt9XsyAUfiq5IZzHGtHkhprwPQtUDPTIcfJM VWU/eZZVDaF+Loscr5rP+rmsyoaGGhSNhodqf5zhHWcV4ul/DWUMD8r0v9bweBLD82l4qHYxPJcK Pg3P51KvwEshzzK8TFYhnvtjSNHwUtwfj8HrHvBVVXb7fJfeNaDhDMTeFS7VUBQFUbyKhoKmomgo CqEaSvVllOwRlAmjKBO0lThyGUeJ4ZjVSzhmtlLg33pWTgsEqtYyXZkX6tbDtAY8wbXaJbdO1eRu jp/6ROIAvMd9CcZAQ9jgmxY2+qZBWZlH+zwtmMICdomYGfZkr2d74gAP5JiGbcJu88hQdnl2ORtC 7WVDFvaT98iQZ/tkb+IAOTYyJGO3DdcYs8+Ojo34gKdqbeXo34aRZ+NI3QE14cwFNeGyusWhblGs CgebKhuwL/dSn9FYhYFurHMCdpayTo4bRRztk6QRROTGc3OzyFyFFOEWGgIbcCu40FgOdmxA24T/ Bwqp0KEKZW5kc3RyZWFtCmVuZG9iago4NCAwIG9iago3MzM1CmVuZG9iago4NSAwIG9iago8PCAv VHlwZSAvRm9udERlc2NyaXB0b3IgL0FzY2VudCA5MDUgL0NhcEhlaWdodCAwIC9EZXNjZW50IC0y MTIgL0ZsYWdzCjk2IC9Gb250QkJveCBbIC01MTggLTMyNSAxMDgyIDk5OCBdIC9Gb250TmFtZSAv SFNETFVBK0FyaWFsLUl0YWxpY01UCi9JdGFsaWNBbmdsZSAtMTIgL1N0ZW1WIDAgL0xlYWRpbmcg MzMgL01heFdpZHRoIDEwOTQgL0ZvbnRGaWxlMiA4MyAwIFIKPj4KZW5kb2JqCjg2IDAgb2JqClsg Mjc4IDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3 NTAgNzUwIDc1MAo3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1 MCA3NTAgNzUwIDc1MCA3NTAgNzUwCjc1MCA3NTAgNzIyIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUw IDc1MCA1NTYgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAKNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAg NzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA1NTYgNTU2IDUwMAo3NTAgNTU2IDI3OCA3 NTAgNTU2IDIyMiA3NTAgNzUwIDc1MCA4MzMgNTU2IDU1NiA3NTAgNzUwIDMzMyA1MDAgMjc4Cjc1 MCA1MDAgNzUwIDc1MCA1MDAgXQplbmRvYmoKNTMgMCBvYmoKPDwgL1R5cGUgL0ZvbnQgL1N1YnR5 cGUgL1RydWVUeXBlIC9CYXNlRm9udCAvSFNETFVBK0FyaWFsLUl0YWxpY01UCi9Gb250RGVzY3Jp cHRvciA4NSAwIFIgL1dpZHRocyA4NiAwIFIgL0ZpcnN0Q2hhciAzMiAvTGFzdENoYXIgMTIxIC9F bmNvZGluZwovTWFjUm9tYW5FbmNvZGluZyA+PgplbmRvYmoKODcgMCBvYmoKPDwgL0xlbmd0aCA4 OCAwIFIgL0xlbmd0aDEgMTM4MzYgL0ZpbHRlciAvRmxhdGVEZWNvZGUgPj4Kc3RyZWFtCnjarXsJ eFNV2vA5597sSXOT3Cw3aZN7k+7pvpGS0tzSBQTZQVpopRQq+8cqKi5UHUULCIqouHzyjzozCkgo oAEcQUWFQcQZEdcZGD9AR+24fOg4A03+99yEQlHnmf95/nt73rO9595z3v29BIQRQkbUhRgkz1gw fRF6CTfCyFtQnp+xfJk4oSlThRC+BSF163WLZi3Y+8mPSxHSfICQavis+TddF6j9+DWETCJCvrdm d06f+flHud8jVPAVrK+aDQOGm8h0hAozoZ85e8GyG2f+0fRP6A+H/sT5C2dMHxJr+Av0H4d+wYLp Ny4if9beB/1D0Bf/a/qCzvWHK05A/3Po5yxauHQZ+iPDIlTEQf/qRUs6F6lfD8NeilphT9thDCN6 HnoiNfo11IXoahghcLqfu1j0//MCOsFbYSNIi3RIjwypnZgApiEzQA5ZkFUZtSEeXv8ZlAeQB2ov 04G8CCVOpsqn8duS8/G+RIK8T6mVKslrItwbFTgRj0rWaCY6jhag+9HDMFaO30bPIhneOhFGGSBL M6pBG9AN6D00KfEtjEroKfQ1KkDVaHYiDvtaieL4VvQUJkAtMwqhd1EnWk9qmCD7JVAwH5cwW/Ad QM8aWPsQcqJj8MT8hB76O0kGqYFVE9ERZpq2IFGS+A4fYA8nOtCvcQ05wT4P0tSL/SyK35lYnXgs 8ThQ4xyT0fdaojSxAFZNQu3oenQL7KAL/Tc6ilvIELI/cS/sqRn2sBK9iI7gIIvYdqDdeMD+FXoE 7UEvo2PoA3QGY2zGubgLv4uPq1DfwfjBxFWJjsRC1IhGo7GoC2YzcBauI1OYKcw25v2+/4mfSnjh 2RPRcnQjuhmtQ+vRFvQ++hB9jBmiJxPJJGYb8qAhaArqAGpugD09iw6jk1iLK/BgLOO78VaynGX6 DoJUscgOFByuUP9+9BjQ9Bm0HR1E76A/wjO/BZoyWMBBPAm34lvxXfg+/CB+Bm/Fz+MviYp8wDDM 7ewb7JfxEwl94tHEs/BeD0pHIsoDzoRAdhego+gLOF8+LsAR/CcSJAUMZo198Xh5YlhiZeL1xPso gHIAdwhqgDOPQpNh1zehO9E+9AasPYreRmfRP4BKDNZjK9BCxAE8Hk/A18MutuGvcR9xAP9CZD7p IceZIHOUncw+37crbo/3xL+OJxJbEtHEa4m3FP5WwXvqgQNtaBFaqnBsN7zndXQa/Q19D+9QYx/s dTgeCed9BJ5/El8AcdKS28hWkmCGMOuZw6zAPhIfHV8QfyS+M1GRGAWyxYDmCKgC7sEgTZNQCzz7 DqDmU+g54MxOkJ4T6O/Yhb24BF+Fr8HNuB3PxgvxIrwY34xvAao+i3fhffgE/hj/nbBETexApyCZ Qe4gG8gucpCcIKcZxExgmpnFzM3MBmYX8w7zOcuxBWwJO4ptZ29iV6iQilE7tG9dcF5Y0NfR92jf a/GieEN8Xnx1/JX4ifinCUNif+IM6HcJ7LEFzYI93grnvxvdh54E+XgO9vhX9Bn6Enj+HdCCwTrs hh37FL7Vw75Hwc4n4xZ8Hdyz8VygfxfegnvwS/gAfgUfxkfwn/An+GuCYfdFcIdBCyaR6+AMj5It JEo+hPt78k8mmylgyphyppZph9OsYu6B8zzMfMKcYQlrZ0vZCexK9k0Vo5qpekj1mOqg6pDqCzWn npqyERMvN1jMW+QVtpaZjzajsYRhviB/IjX4VnIe/5Zk4FfgbRnMWGYsqSdhRPA+kPIFiNc8ppbU EuERp2mnzyCbSCEzmc1mjGgZ6BsiU8jdpB39Br+EzpPhIGnLmaNkM5nGPMY+wNbi99FKeCciJvwD qkN1uBZ49y5aDBwqZLazbytmVMtcUC0gpsQq9jMVYf4EdnAIJswf8BTci8cSB1ArTO5DAehzuBfq q0ADPwTJ34MnoxB7illDRpCPYWw+2oBfgTPuQ/PJPvxr4EsI9HEJHosfZ0rRbXgxUKMazSUPIj9Z RPwgz5PQ/+I7sB009zzwJpNch1jGRGag46QFuP4OtpIifBvI6QK0GnejAtyHD6C3yP2oCncyL18Q +nIJvtCLdzDD0Q58nj3MHiYsPOkVoGYJWA8ZJOQpsBGTQDMlJhukJoRUpADkvw0s4NXIQr7Ht5D5 aA5+hPkbfobUoTGok1lKmvBD8e/ZOqYcKLYXrEm9ulqLVDWqDLYCOP4ZqgVpnAWuZzZ7UnUHbauf 7dsky3KkdkhNeHB1aFBlRXlZaUlxUWFBMD8vNyc7KzPgl0SfNyPd4xZcToedt1ktnDnNZDTodVqN WsUyBKOCxkBTuxjNbo+y2YHhwwtpPzAdBqZfNtAeFWGoaSBOVGxX0MSBmDJgXncFppzElPsxMSfW oJrCArExIEaPNgTEGJ4yrhnaaxsCLWK0V2mPUtrrlbYJ2pIEC8RG1+wGMYrbxcZo0/LZ3Y3tDfC4 HQZ9faC+U19YgHboDdA0QCvqDCzagZ21WGkQZ+PgHQRpTbCpqDvQ0BgVAg10B1Emq3H6zOjYcc2N DR5JaiksiOL6GYGOKAoMjZqDCgqqV14TVddHNcprxDn0NGi1uKPgQPeaGIc62oPGmYGZ01ubo8z0 FvoOSxDe2xB1rjjtutSFh1vrm1ddPuthuhtdc0Ta7e5eJUY3j2u+fFaisKUFngFrSVZTe3cTvHoN EHHkBBHeRu5qaY7iu+CVIj0JPVXyfJ2BRjrSPleM6gJDA7O757YDa9zdUTT+JqnH7Zb3JE4hd6PY PbE5IEUjnkDL9Ib0HTzqHn/TTkEWhYEzhQU7OEuSsDvSzKmG0XR5o7N/Tmkp6LQ1cnw/ZTHdUeAq EIioOEOEnTQH4EwhCjpDqHtGCNDgasGwKjoTODInqqtv7+YG03G6PqrK4gJi9/cIJCDQ+9XAkemp EXUW9z2iTSon/aIG8xfb0WAwmp9PRURTDzyFPdYq/crCguUxMiewiBOhAvKhsUDb6S2Di4H8kkQZ vDomow7oRLvGNSf7Iurw9CC5ONgSJe105sDFGfskOtN1caZ/eXsAJHmXEsPao9rs/j8z57A1zh4c xY5/M92ZnB85ITBy3JRmsbG7PUXbkRMH9JLzof65VCtqq29mPCTVIh5GmQWhbO1Hpp1mY5TNgj+1 ItQzYxotSKUygsWmKNc+PAlb9JL0Hy6KJb6hq5Tq0rLUNqODgwP74QH9AdszdjOwYTabjJw4pbtb P2CuCSxQd3dTQGzqbu+eHkt0dQRELtC9B/xpdveixvaLHI0l9q72RJvWtMAhZuPBheD5lNAegnsG IvumHWpNDBt3waiKpQ0G6dUqaLzAMMSt09CxFzAStGNudgVHc+dqRvXVjOZ+qBnF9dWgSE1fDS2l JZJFsmQBwBBAXhCZAxdkFTqPRPYAzWD2Eg1rIyvhfW7ZiA4Q5FYRgZ2xhT7wNHcWFY/qLS3BdqmS tV34LVl5440I46OJTyHI/xZyjXRZj3u0BvYDg5C2YA/2ImUfo3pRBFZllZeBtVcHAv7syooqMikz NHbcIAq+HRMaPJoWeP/ZxGTmC9UCyFYWyIN1OgcWdEwIVeua8FW6qbp5uuX4Rt292nt1D+FNumfw s7oX0Av4TXxYdwKfxX/T/YB/1DkNOmyI4UO7GUMtmqqL4R7Y1FTt74sZzLxvieF9O15yBblzbX29 53pPo2J6nMVtbRA/KvsahKuqlH0yp/paLR6LoCdPGfg0i6DK/FdzlmA22lW/c6YJZgPwRjk35E4m iJtLZZ25x2HQ9iC1dR92QEzJYsdug0EQ0i8RAjjRy1FiUHLgAeSw/RJxLgGmY+yg8Gha+tb1UwyC HrQ+cZK9hulCuRALdMjjntM87XuuiMnWZPnC7DLbDe7lni7+LvcD/Eb3Fs1m/mn388W7NS+l7eB3 ufd4j6SdK7XrwbHnY+ZRy4NucnNRd9FjRc+lbSl6vfS90jOl2lx/jDwvu7OKpawsv+TPtWbYnHlV EqrKw0y5UVdQFcOn5Cn4nlykL5cYg05CBVzBogKmIC9sNObyj3NShoZOmJAoSrLJETFLuFiKSGOk adKT0nZpv3RS0krukHNdiaSm8wvVT6r3q0+qWbUwKH+fK4YL7qLEa8PBUX1nQZ7bFuMgFW2Fjr3W 6uLe4jZoRWrO9Vqs1VZnNYaq2lptqUZc3xCYd2Hu+3MHS0vQyKgwYWQ0EzR5P9IkfkQViW9QJRQh cW6nVVukDSlXC2pbDKgGQOUBdR/yAootcYDOLG7DbVJlUlSonDgrK7IDfk1KdqoGKbJjSfGSyVbm 7LyjvKxqENP84jsPP3fq/cH3jOnq6tgh6jinPm3G42Of7Fnkc7t9r4d/ddWLs0bfsGTBvhk3Pbpp 4YoXzNw9jddV611Wi97szn9iRt9xp83qwr+2cGPC46+ePXka1dctwPuRwPt89M1uSW8wR+yxxA9y ATTetH+S9WHOKd8p6cusL3I0mfYcR4M4KmtUziSxLWtKzlzzXGFO1r2C0RFLfCcvtfEttmvs87Ku y/nBrVK7Bc7uzuPyrFnubu4x7iHXRvcz9mcAN5BttZgF3gOJkzZNSHeaTYixGNA9FilPY9jJqtN/ 7ZQChrSwtmWzD6/3HfARn7uAl7Jlsy6yORubs33Z68HoCcGD911iLPBzVK/C2FHnqHZQZvaetgAn gZGY1sBPC+0ADxe3hUKgsDgYxEkW2O1qSv4cSuyLtE6qMGULqqxA5WXM6y6rzYWdNouTqLc/uO/V E891HBlv5yzOzqcOHYmfx4YjrzCmdMqHl31up2dY1xcPP3V8+FjeaQkOnYeZN49go5JwoC6gdzHQ W4A8+R15jt6xyU7KyFAyHjK+N8gbtj8IH1k/Ej7x/I/rjO9fDpOQnp9eQULeEZ6rfa2eKb6Fnvm+ 2zxrPJvSN3lfVJmvd+xNP8gctB5OP+xVa1+3uEUR7KklQ3JqWMliME50hzcjvAj4HMNnZKdfDOPw Zh4v5Pfzx/iTPMsLUv7Wywg5qreXkhHsGiUhiD4YfK5XkXxjSvJ7HLwavN0uD+/zkljiq36Rx/An OQaQr59+SKNQV8MWXvid48yz175dZ0vjXFzJ97d/ED+JzYfexvrJwnsbNhx34yeeerO23CxYLFzZ ZOw5/CJWx//39tXPb12LlC9k70NGNQXyhgp0RM6SjWNVXao7jbeXbjb2GHcFXw0eD+qdWrPOeIjj /LqKIlSKS2OEfQEhfxHRgquTZTfGfm1mrh9lteVJGQhZRaGo0KXWafX+GF4j66sgNRLdxzjMxfBG 2VRsl+2L7O/YWbtQef0e/FbSELeNOtcGpKkBhxYBr1hDfWPfacVuIK4XzMXxmmTddrFxHBxn/U1y Wn7QAywt8KGgJ8+HURAHb78dt2GLlFR5KWkYyi0pC1GuyKjToUz6cxQRtadwSTFe7HW7vX0LKTzy AoUvbL3vhlXldhevtT08+79uwPfSQcbUN4yqPsiwk+xxQWvl3McdWofV6mSc8xtX0hEqm7fGb2Nv A9nMQeXYK5c28ot48on0btZX0ums89K5TPW8vAWFM4pnlK8w3ZK3uHxNXlf5E3n3l2/J21y+15tG tNTOdyAVVhXrVCqtzk+QN1jqEjmnCNxM824olUR9UEIbsjXaMFFjNc7NELGo13O6zbqojjHrxuim 6bbrjulUOndlkdQVWB/YHIgG2P2BY4FTgW8CbECoyJ8+QFwVraduEZiBIiC0p6nqR2qon6TcuGTA 21IW/KIc70OexDnkTpzrydeWxRI/9ni1KAa9Am0JrfKM5XSw0FEcS3wRuuxK2nBcedHfWnhNGkm5 3XInFf5Kat9JZYW1vMxWccmoMLe/TpmxONO1qHXU32nz2xE35DhWvbft/Plt7606snbtH/6wdu0R cuhRyo8LeyYOLbg212kDs3P1Vfl1F/ZgvHs3RvGRD751dMODR4+CLkwCXVgAuhDCk+XCTe7zImGx Hc9UX69ejx8km/HTJIp3Ev0z6t9odql2a97QfKA56da4tRYn5RRn5n084VtdPO90+S15xXTQUNBa UlBQXOLP4/S0n4ZM2NSqM5n0Oj+X9OSGrNaUJw+V0X6gsri0srKs1B/CYl66xObl5gK7Q4jVcHqt ThROurArRp6SDYORJJbuLzlWQkpi+Mud1cNSrAQWUjOjaBTwMMLVKHev5XL2nTuIfqHT9m+nFK3j 3B6VRp3lUQk+7NakJ/UOFA/EwZoShz1InTi3WzT6eK9DYTNuo4y2KNHcJZfcr59Jt635yXhKEPD4 sRumdtzbeq1PEHzxr6ljuPbO61vriudPo5yPK/yfpqgv+1nf+cnDGteN6ftHv5IyU1cUijf0fXVx gK29qKO/B547VBaIr9PRSjnfL5QJsjBemCEsE34laGwmrpnn/Sa1UdesUvmNjnRho93uT2deJzH8 4AvpapNRj/A+PA3WEwi70lhWJdrH8JgXMsatTJo2GhpBwK+E/JEfegeYs8utGSiBPVBpu8JySRcp QNbfshKPoAfvc9Fj4hHfez1un8ry4YfxcRe+u8wegWdN2vZ9cDYe5NmF2uSKDvtS+512MArGZmrL wXo3U8ttddk3Wix+FwKDjbBo4bgx3H6O4QTh8v3Trf+bff/inu8fuOPv6I7Zz84vuJIPGNlhr/vB VjYRtVxjHmQOpVWbB5trzEPMsrne3KizZhurjLs8PQVsDq7CZFJ6h6YjfZlmWbqqSlOW3qhpTJ+k UZVoBw1RtOzkYDy4qXbw4CG1/kF2Mx3yilY81vqO9ZT1GyuLrJxVtjLWpjSr1Zzmt2f5FAVGfs5P /E1ev9/n9WdVlSQHy7lyUt5UXF5eUuyvapLpYOfJelzfFKmvlyP+wmK1N7uoMDcjXY01+YPkMGpS 50uMW9LpGM2gqqqsLLvelCY6HbKvssTR5SCOC9kZXjEnm/azu7JJ9oVaVCxGammojWr31x6rZWqF YfnbXKkgW5Gfc23Bmv4KuEE1OmmVIzW9llSMba1Gv6DF/7bXtvhnlVzMzXMJeiOrMmTlsTk+rFIL eqcP56ryfdhldFOlB63naqjig+q3tYH2e1LaX6dH+sTfEQtFk/gITMFHCCfevWjv8eL6VjAQmsSB nQ53LUQ/So2h7oEaHhsEa2GzK/F8leKu++1FQDEUl+xE4ErDEbjCcHw+b35dhxRaOnhq1bBhVAof G11edF1dk9IcU1pYMKReGf6UgiQG0zFpaWNTU2P46il9u6mkkofliY2dfe8q7fvrJ2fkzUx2LgUA Six6P8SiNSDDBuTEw+WQ1cE6eKeDOYwPG94jH6v+rHnPoJ6nmWMhnaSTnaOdo59rmm/ptF3n1Nol xizpIFnTGCVE6WEWIkqd5lRq2WSvjCLMoRLUDqYmRlbJLitkZ4CmlgFnISRox9Sn1N+oVeoY/nSn C8Tnoi8Ay9zb17aYWmgI5iNUZhTHbeh33A5w3Hzi3C6OT+OdexOfQnL16U6T1+K95KPbIMRHlG2y wcFznghPgYWmNjazN2LgAWj1ADQUwPhXcobVENHwBitMAnDwFmctT4GNN/MU46BshYZeb+RgJQDC mH01OEi5f/nVgnl0Keq9zP2zNfHeVw/G/46tB1/Ftkl/3bz5r7Tg7Qfi32DL/gPYEv/mlf/+y8kn Hj91EnhTCLHYZOBNFoSwhXKkVG+uzoFSWTgOTyJtppkYeKKeZ1qGb85fUmR4TX1A/6HmQ91HOR+W nlWf0WsFpoC5WbOG2cRsZdSOdMVvC8UZgpCe4XckLYzBemiAOanzF6csCTblFZvD9vQwOL60Ysmg z5PwBlaDfOEsdbZk1mKtu7wApYlec8aYjGkZCzPYDKHs2ssSMmoC6JejNuXLTY0SlP1cTPbLnvuK YK0n11iyF7heCFwPGk0iplwvTfx5R06gn+cKx2laZ0/5bMqInF900JexBueP3Hr9LX9aGu/7/V/X vKX4ZyWkXqz4Z+aJdx/ZdPz4poePMx2bprYuO7ZkdzzxYlxNFQrCMycbpgoVn3P/sXfW3//OMZpT 0xyFuQEF0b9k/pY0XKAbo59rvcl6r/Uh9RM2TXoyjPIdCvh8/oA/3WPfS55HLshNdLzLZef9nmAW xRiTOzozNzcr0x80pPHKv26oNCZsQ3wap8/MCqOgWh/hJFZjD3v84fR0j96s+UZDNO5CxIuZ5sDY QDKA/iagDggFffdd8o6jubNtYJOVj0k0WwYTDXyhrEla5upq9P/AI0uKR7ttfJrDmn7RZKbYoOQt qeRa80tZDiFPP9U48nbBpk+zBSqEQY/tx8sUP7zA6xZ8Rx6jkOk4/uCkTrcNgpyAu3lLvEKhvxUS 8pcu+uRjiZNMHHSmAX8r38NH0uuI9WrUguY0bBW3Dvo/obdsh4f+xXbCcaL246Ff2k5XfD70gu1c xY9DrQab2qGq1Q312ewOe61n6Gr/xop9ZsNk25TQnNDc8IrQbeF7Q/eGn+F7eP194d0+Mk4bzAtk l8pDaircLnOaxm6sRhVlJQG2qMqcZmT0iLEI4SFDJItUr4/hyl2MWISLYvghOT27SpJQWDOpWhrj neZd6GW87qbSiYFwnl2Sqd10gIWUWxbm4TyhsV7DqLP1kuHarqRm0dg4gpPZJw72cn2nFRXr6+tF lJNtAPos1cWpj1j9Tlb5ClJNv32AkwwNGmoV07NsWc5auw+FPdU+PEgEYB0KXUfE5UNOV+2QwRk1 Puxxh2tCviof4ussivOkPjwJcNLWKQH0RfbvCvMV+vSXEp8hZ+Ir1JD4qqeWHwSGdaffUZN+KXvC bYshrlb8aQisrg6CiDAPIERtsIuzQw9AAzW6DTyY2QbeYI6k0+cAZSjSi9TV8BRcZnTB3qe+5aTE LSf15czpoHcqe+aVDz2Vqc8+MEeVv7IiJzszO/k5jrmFBqMu5StPaPyqtaPDTSV3b2+YPu3tN99c qbWbqMJbBWdg08KnN48bH3/znquPb3ieCWaAqK73uh1CTU6oOlhZk5tutrkCtwyf99tOP5/m9m4D +bUX+UoiKxpGFxeLFbNr5q+k8voA+N8w+wAqQIflzPMebPK4PeRp/W79q/p39af1quVpd6dtTPtN 2huGEwa1U4s11E6weIls17KsRuvHHK+zW8ycxcqrBGNeDD8lW7zhzExNGGOkNkqCgb+HjeFnZb6g ADKxbOkNlM6li+mL0venq8AnnNlZSMM2EKLTyseMc8pnH7DTEP/3Jj+bUQm6QuXpVwy3R28wuHU+ pPcYfSj5FYOmS/B3UcUt/JWfgrIrB37VcNjxXnxIMbeh6xdPemMQb+JcJvEfizc8/5gS7lBmMB1U u/v+eFVHuWgSLGaTNKr7elJMB3+kSJSOoUSCfZDpQLlMgfxMriPHeTfznOMZSHL3OHY5tYhwZKVj nWO742XHSUfcod1MouQYYbSs1u5iXfZcksfm2nOcITZkH84Ot09mJ/PN9mahOfc6PI+dbZ/lnCXM yr2ZvdH+iOMh52/IFvZ39s3O3WQfG7NHnS8KL+Yedrzp/Nhx3Pk3x2ln0ODwOIIk6Ag6Vwmrcrc6 9jneUL3Bf+L4HH/u/JGcd/zotJhsStDPccU8x9kgXePtUg4dKliUiVGmmClnMt/Q1ubMdzKZRZld mYTLHJtJMjM35WZm5uT6pVxkVNMFedN0K3Xr6KcTn26Mjvlah7fr9utO0gGs021S6XRqyAJVrOim 2JaMjGIhI8Mt+EXBtZE4nGIsMUIus7OMyKtYVrTzPChQLkKiSwBPJBBMGCy6nNCGlJRgRrQ7AMNB 9uNTEDIuA591iqZCkEgGWDQRY2Yiq88JS+6waAub1GGjJIomk1G90IVdrwmYfl7LRhsEuaRSkHOD FYKclQMgwwtAcAMwWyqEsNyei3P34d9CDOTEq2WnYxKRS6srCMUjFI/InKUC0tnfyiaV2G7H9td4 dgMfVtFwvKSSVjtD1RVKN5jswmuUGp6g1LBeqeFhtJatDmeFSrZXrlStUxGkGqMiqt/jT1EezsIS nHGmEpa29Vub3tMCd7rNzfXRTp/rrMD1tbldvcnJc2fpJHJFUh8xknfkXA13mjb6elepioLaW7mD ULsuNWgN1apbDwaDLqTt17hvi8HlXt5tg/Rn8U/HfjpIVRVfipp352oFLctdMsF4iaRhmBzmokqm 0hGbrdxmu2KMuWf2ntjs5/Oovn1GwbyNO2fG1s2l2fFZ6qdzMUnvO40ZE1VLRV+vI3zfV+TRi/1k ft8JcdFc0NN6skLe6LP4rMQasky2EA/NFHz+drzAulBaGGivfw2/xr1tfVt6K/BW2asVr9abtciF HvEzqAxb6y3W+gDnD3BSRXkZlirKApyVE3EZj3FZRb3VahWlCl6SKkgYh81hLqy3ha1hKSyG3aXh snBmOBDOHxquD1eGK8Jhub4+EgpFAoGcoqKcSIuqIoaLdon1j0Y4+m9SHoxVRklyGI0q5MAORwZ+ 1KxaCMLhbiyD+Z2BR3OsCp70aE6LOaM4FQyrMoQGvd6tz1eH1Wf3Yg1I0EDnfVo45+oVOADUfwuj TruAZ23guQUoFso/mD3t7nVxp+kgHUjVbuTieuG6AqhWFQUVj2pN/GGnUBaxxhLHdzoLab11J59L 6x92WgO0/pRmaFD/pcdTU5tyniB0VPZoshTgqmA9VwCLORlWcnpYxnlhDecFJ835+1cpy8xwKf/8 idqSKXJ54lNZB4mUxQs5VDm8Th4BDYve4ay16C3W2vo6rzWCKagflG6JYArqB3k4aAGo5wVzBFMg 6TPE2gozgDJe8NRyNCYoo0EA1NZUXQ852U6Op3n4QdkEjUANAIkCHPzJhfoDBVzmGBCEYjwgRYDU APuzB8g/xgMUQh0gm/Ht2bzZ7Yt/R8V/dXxPfN8axZF97XWbbdn49vhzmTaYP0M//83EHpwxkyrL GTqbiV+Pr9M4TMl0HFfH30ym5iaHBqLm4VplhmYUX2NLUn+MDq2iP5Czd6HvxH9+Ms1c871Wp1V+ B/fyhKi6v7bHb1Mh1QLo6lK/JVbW0d9Uge4dTZxJnGE/75+5eNnZo2gWlL3sUXwU6rOpej2ULVC6 oLwP5VYok6D8Hso+KHTd/VAKU3jHoDwAJQQFdB2F0Hh0I3obW8EDH2Jmsw7276rP1deot2rTtJ/o 9uqXG942LklrMp9N/igEzUMaNBWxYPk55EB1cNw5mu3Qw8oprKl9q+m3t6EN48aNbwjWL7x+yZzO JaM7bxg7YdRE+k1DueKBgb8VvHTS/t8xZ6AG1ISGo6vQCDQSXY3GoLGw2wloEroGTUZT0NS6OrSf KYR3blfgZqYA4DoFTlPgGAWuVGZXKu2FSnuh0o4o7UiqTWHxZdCnQDOFjETWIB5GxFTtI6uRF2ov Wd3j9cl1OrILjYXSDmURlC6yq0dlNdfxgEdxi6GMgTINykooT0LZDkWLIskZ2UAiJMKMIWMYVo4x eTtrasqUurwqWadnJGuju8xct4TJg+PkoSehMLDFPNhyHhzhYs8HhQBRcoA8OegYlJNQWJjLAZLk AHFygEw5sD5HwVIreF9DSUBhgDw58PyBOCpltQ9K8WVPoaO5MJILvVxYkwu4uTB6EiBWVtD5sVDW QdmfmvMrbPIrZPfDs/yw22KAEaVlBuhj/D1EZ44BffFgc10l0H0MFJgka4Gaa4Fua2GjlK7AL5iJ pDDWQdkORcXkwZ0Ddy7cfrgluEW4gXsM5HRkPdzr4L4P7rVwr4F7NXCC3x7cHyTTKhdWrqxcV/lk 5fbK/ZWafWQ63O2kXdYjhwMk1GrRuus4wqJWZML/UuA2BS5RoKxAp+xuNZ1uNR1qNW1qNW1sNTW3 mka3mppaTcWtphjukJ1B08dB0/qg6ZqgqSpoqgyayoOmvKCpzoJb8GRkQi8rcKgCyxToV2AGntxj QrqX8FQkaUFWcc4u6XbfGSnG4h7fnVJMC9Udyd7UZBWmgy/4SqRZvoLkSHayykxWkwbtxVuRBgfl As1hzTSNrKnWFGkKNbmaHE1A49PwWquWA6Ng1Oq1Wq0aInOiRVo+ljglB6nq82qOVmqWQlZpcwQp /zdCsW8EawnocdTGjCQjJwyFsOfADDSyQ4z+MCEQw/pxU6KqwFActY5EIycOdUUHBUfGNInx0VBw ZFQ3dmrzDozva4FelNwTw2hicwwn6NBdHvqjxj0Ig/9e60nVLS10TfMOFq9d24IcyyOuiLXWUt3U 8DOgPQUvc0Ou4IBORvShkROao89ltETLaCOR0TISKEZ/AbmHhEhVY8MeMohWLc179F0k1Diejuu7 Glou4SERxhv2IIlWCh4SKR4Sr8DzkkEUL4tWSTyvgucdgLdjiNTYsEOSLuIMUXCGDMSZNRBnloIz K4XDJHGky3A0p5Ck4EiaUz/B8f4HOFk/i3MZLTuHBv/NhfegEfjEjvoV9Oej7YHGTijt0dXLZ7ui XR2iuAfV4xOpX5Zmt3fMmE3r6Z0xfCLQ2RCtDzSIO0as+Ol8dAWdHhFo2IFWNE5s3rFC7mzoGSGP aAxMb2jZOWx6/rYBr7v34ut25E//mYdNpw/Lp+8atu1nprfR6WH0Xdvou7bRdw2ThynvUmQehFKL hrZAHKbUO4lBD+Lb7pFahjq4RbWKLIcl122evSzCv0OGYEvUGBgaNUGhU4V1hXV0CnSMTqXRXwmn ply3hSXPXvy71BQHw5bAUORqnNMAf0uXphr/yd/SpUuXXbv02qW0Vv6WLrseihKhLUVLlyHYfp1R 8Ws+sMLUJq+GskaxzczSpS3LUBL3ygv9X+CAj/cKZW5kc3RyZWFtCmVuZG9iago4OCAwIG9iago5 NTUzCmVuZG9iago4OSAwIG9iago8PCAvVHlwZSAvRm9udERlc2NyaXB0b3IgL0FzY2VudCA4MzMg L0NhcEhlaWdodCAwIC9EZXNjZW50IC0zMDAgL0ZsYWdzCjMzIC9Gb250QkJveCBbIC0xNCAtMjc0 IDYzOCA3ODMgXSAvRm9udE5hbWUgL0JEUVFSRCtDb3VyaWVyTmV3UFNNVAovSXRhbGljQW5nbGUg MCAvU3RlbVYgMCAvTWF4V2lkdGggNjAwIC9Gb250RmlsZTIgODcgMCBSID4+CmVuZG9iago5MCAw IG9iagpbIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYw MCA2MDAgNjAwIDYwMCA2MDAKNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAw IDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMAo2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAg NjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwCjYwMCA2MDAgNjAwIDYwMCA2 MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAKNjAwIDYw MCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCBdCmVuZG9iagoxOSAwIG9iago8PCAvVHlwZSAvRm9u dCAvU3VidHlwZSAvVHJ1ZVR5cGUgL0Jhc2VGb250IC9CRFFRUkQrQ291cmllck5ld1BTTVQKL0Zv bnREZXNjcmlwdG9yIDg5IDAgUiAvV2lkdGhzIDkwIDAgUiAvRmlyc3RDaGFyIDQ1IC9MYXN0Q2hh ciAxMjAgL0VuY29kaW5nCi9NYWNSb21hbkVuY29kaW5nID4+CmVuZG9iago5MSAwIG9iago8PCAv TGVuZ3RoIDkyIDAgUiAvTGVuZ3RoMSAzNjYxNiAvRmlsdGVyIC9GbGF0ZURlY29kZSA+PgpzdHJl YW0KeNqkvAl8VEXWN1xVd+2+vdze13S600lnaUIC2SGaqxAQISyyJUhLkB1EEjYFQYOyGVGjM+Iu uIOKBAgYlnmMDKPjwoC74ijMTFR0jDIOMo6QznequjsE53m/b97fl07fqrv0vVV1zvmfpU5dhBFC BtSEOKTNWDi9wXngp3o48g5C2Dpj+dLg5ob3lkP9FEJi4eyGOQtrn/7HFQjJKkLCu3NuWDF70/DM 6QgZOxFa/N3cWdNn/n303m6Emo7DPUrnwgHrQsO7CK2RYT9z7sKlN088EYjAfj+4Z+UNi2ZM//a0 shah2+fC+a0Lp9/cIOgNDyB0B70+eOP0hbNOOWrnwH4O3W9YtGSprNM1w/5VsN/RsHhWwzXxI98g tBau1/fAMYxof2iPEBbR/+cfJhxseUGUoE86vWIwmsyqxWqzO5z0tMvt8fr8aYH0YCgDhTOzItk5 uXnRfvn9C1I3KBwwsKi4pLSsvGLQ4MrLLq/SrrhyyNDqYcOvGoH+67+r/4trRiaKUahmNPq/+hMO IA98vcLzyMNHkBuhnq/he5qW8Xk9p+l5WpJv4eL25BehbWgHnod2oFfRYXwGfrUT7Udt6I/IhYai x9Aq9Fu0AYloChy5E10DHwGO/xZ7etpQAXoSeOlJdBSunYxuRQeQE7t7vkG3oXXc+/CrdciIMtAV aCxahO7Go3qWoanoJH8HKoPO3YgacFNPbc89Pff3PIOeRfu5P/Z0IwV50Qz4HO35Xvik588oH37x AHoYncT36/YiDZ7SBFc+jhajR7gYj3vm9PwCLQihm6ANPKpBR3EHicLdZ6GvsRuv4obAXZ7uae05 Alf5UQzNRY+gA7gEDychYWpPTc9R5IRn3Ax3fRjtRvvg045+h05gg3Cm55meM8iD+qER0J829Cfc wcW718Sr6EDDKOWiCjizCP0PegMdx2H8GlkkGISBgias7PkA2dEANBFa+zz88iv8L3IrfG7jXueH 9VyJTDAu99HRRn9Af8FeXIDH4EkklywiT3CLkQxPHACfmWgejPdDcPcvcBTvIwZyjHuaf5E/L6bF T/WYgCIR9Ch6HL2GjdDTIF6Cb8cf4b+RIWQaeZT8lfstv51/T5oOvb4OLUR3oxfRv7AVl+Nx+Fo8 F6/CG/B9+GF8FB/Hp8kVZAJZQH7g5nKN3O/4K+Eznl/C3yGsF+4ST8dr40fi78b/1TOwZz0aB/yw Blr/AHoCerYfHUOfwuck+isWsIJN8AniEJ6Ib4HPrfhu/BTehrfjNnjKcfxX/A3+Ef+EzxMEH5H4 SIhkwCdMFpObyG/JY+QYfI6T78i/OReXwUW5Eq6Sq+MWQas2cC3w2cv9hffyx/geGOeBwmZhi7BN eFE4LJwRDdLtMpLfufB0d173F3EU3xjfHN8db+v5C3IADb0wCumoElo/HT7zgd6bgeN2ovexAcbO i/Pw5XgUjMw0PB834pthJNfiR/CzrO0v40MwSh/jH6DNRuJnbe5PSsiVZAx8riOzSCNpIfeTNvIR +YWTOIUzcw4ujxvOxbhZ3FJuBbeZa+Xe4T7n/sqd4y7Ap4fX8+l8Bh/ho/xwfhq/jH+C/5r/Wpgq vC18KerFheJ6sV38h1QqXS6NlcZJMeleaZ/0gVwP3Pl7tBe9cgmqneLWcNXcXnQPKeI95E/kT8DP 09BMroYAp5JteCNZjdtIpnCzOJgMxqPRGT4CY/062ULOkcFcDR6Jx6P5ZEDibqKdfwGKSv73qIs/ BH37E9z5ZtGAbyU/iAa0GyNSAc/8A1fIR7m30QnuJJb4J9FnvB67cBd5nhsLXPA7/nKhFoW4x9DL XCNejfaSakDs8/Im4OPR+AXAhQl4IP6Z60EcGQ1cVMb9Dd2BFpBPUBfI8Ub0IJ7Jz0H3oCK8Cn2N ngOpyBVuFPNEB36TzOObiQ23IcJvh95V4EzMCXa0Fse4R8QfyKdoGTrG69EX3EvQ+mPkZa6GPyNc g+eCBKxG61Fjzxq0Qqjl38NzEIcnoSz+FKDbKm4gH4LyNkCVqYBp+0C6DwAOXMHVwBE3cM4o4IuJ gBCPwOchwAkeOGgeyPhkQLE/oTZxAmlHcwQTBtQBDfN2/Bo0pec59HDPHHRjz/0oH/BgQ88quOM2 9CW6F23D6+K3oAYUAMn5Ao8ShpFjwrCefNJMPiXjyeZL6QujnYXd6Fv4vIyGocuFg6iZ/xiNR1U9 m3o+BO7OAYR9GF0PmqUTevk9POEqrgMVxUeTXT3DuAbo70k0ruf5nnSsR3N7bkBj0CH0rCSg6VJU GzJxwhVa1eWXVQ4eVFFeVlJcNHBAYUH//H7RvNyc7EhWZjgjFEwPpPl9Xo/b5XTYbVaLajYZDYpe J0uiwHMEo37V4WH1wdZIfSsfCV91VT7dD0+HA9P7HKhvDcKhYZde0xqsZ5cFL71Sgytn/+pKLXGl 1nslVoOVqDK/X7A6HGw9OjQcbMdTxtVC/e6h4bpgaxer17B6C6sboR4KwQ+C1e65Q4OtuD5Y3Tps +dzm6vqhcLtdin5IeMgsfX4/tEuvQFWBWqsr3LALuy7HrEJc1YN2ESQboVGt3vDQ6lZPeChtQSuX VT19ZuvYcbXVQ32hUF1+v1Y8ZEb4+lYUvrLVHGWXoCHsMa3ikFaJPSY4j/YG3RXc1a+jeVO7iq6v jxpmhmdOn1rbyk2vo8+wROG5Q1tdKzvdF3fh5tYhtRv6nvVxzdXueUG629y8Idi6dVxt37Mhuq2r g3vAb0nWsPrmYfDoTTCII8cH4WlkXV1tK14HjwzSntBeJfo3K1xNj9TPD7bqwleG5zbPrwfSeJtb 0TUrQru9Xm1/zynkrQ42T6gNh1qrfOG66UP9u+yo+ZoVezxa0HPpmfx+u1RLYmB3mczJisHYtzKr 9xyrsctpbeQ1vSOLaYvCI4AhWoMzgtCS2jD0qZxuZpWj5hnlcBn81WH4VetMoMi8Vt2Q+mZ1ED1O f98qZKnhYPNPCDgg3PXdpUemJ4+IWepPiFYpn/SyGpxP1Vuj0da8PMoi0hCgKbTxcrZfkt9veTsJ hxvUIBQwfGgsjO30ukEFMPyhECXwXe0auh52WpvG1Sb2g+h6326kFUTrWkk9PdOROuOYSM80pc70 /rw+DJzcxsxfR6sc6f03q05b9dxBrdj5/3J6VuL8yPHhkeOm1Aarm+uTYztywiV7ifPlveeStVbb kFrOR5I14uPYWWDKqb0X051aQyufBf8iY+qZ7ZIMXMmO4OCwVrX+qsS2Th8K/Zc/au85Q3/Fios/ SzazdVD00v3Bl+xf0jxDMwcNBjU4csKU5mb9JeeA1RIPHJEsgOPRhNpQcEgrmgiSmQX/7T0d5fRb 52vVYMiG0AuA/xKHkruXXOhL1uvgj3Jnfr9hAHTNzcPCwWHN9c3T23uarg8H1XDzfnKYHG5uqK5P MU57z4G7fK3DNtXBWM3Fg0AoCLqSivGQCbV9CcS4vi6fqguCmYEqILB4JXRlG8GdotROHtZsSOA7 OaSX+E6MPLIodBLuECh+HZiB/ZE7qp6r7K4crZ6trOmuRFVQVy/AZkBhyBKyZMEGg9q7EOQ6LmgC Oo+CfAc8C7WCfroX/AoB6dDqXSJt126ChHayU1PkSlGvG8RXioMwLujs7kRV3V9V+Xb52dkInCVI 1Ctvc7pBQjlficrhOq6SkCDG+G29XlkTevIhdzQKLYpV1qhdaifcolP9HlVV1ajdX40cX7tH4BHG aqVaWVc3oNDGWYosHFdS5Pi67GTx08fwDZwOV8cPXvhX/LdHj9K2XsftITextipo2X5win7ek5FV LLT3/KxlRHKLFVEPapHHSBBE5XudLHMcQZJcqTfrmnREBwTUHEZzse4LzPGVBGtGSzH2GBqfd9Mm Rumoqd3RWCUbPNqo7krYYIu1ooJ+BxTiaNRGm8cVsW3LwKP5nw84Wsjtwa4zZ+LfJLaUarU9XwjZ wvtgd/RDpfgy7Y8rHYudi10r+68sWO98ruBzJG9Oe9pJ7iy4o5Tc4V8bIm1OXO+aHiJOh+acj7gX AiecZIl/SRpZ5l3sI8vQLU7S7LrDR7Y7XnaSOwLNQdKsv8NP3g6+nk2OOg/7yAHv63Yyr/SAk8xz zSoiswrwpKKppWRY0ZR0UuO80kcKvRXpJOLLDBKUnx/I76/XI5/TmeYIOp3B4AF9vl2vz4/kqrg4 NzCIU3zr08LX1dsabFttXIFNsxHbn9PudWN3O5mi+T2XBxYH03BaeXnudVuN2Lh1wHVBCUvzyxof ojwY6zob6wKSd57tikEBdeCbzq6qrg2m/lHTavWIZKrcYKKFWskqMLCx//xDyTJLFMMZ2ZGS4tKy CN0WDQQzRsClZS5RcrqkCC4tLSmOhDNEh93pwlikZdHAUu5o3Xsr/7J2wc6XZ1x57PHNr8b/jqV8 z8HCa2Y1rVgYDyyrnjZ8xPRwGNfE990/+57bx+3YMWPGQ6se3vjZ+MX3XLn29+1r3v1tfFft0pyO VeuvvXcYt656btXIadcNzRiZ112CH578wIi6jlksgjEFyJ/NeDKiOZDAYeF7grg1QdyCCZ4vUu6i Y4KqunCCixIstLE/YxzrTz/Fv2f3WRUfR+qBZ1R0mabPNmOkWiVZVdtx0R60xSRDqVmkLabrEKdy QY7jXrI8vonduvtcl3oO7l9ZVUlHEkeIpbistKxIlODjUDE++cCfaqYcWrMi+7JwFEfj4w7hn7Hp +xPd54/XNW8++Lt4ejx4yfNnaYYckqMSnV7FyKqjLdBv4TCUbWgLd50J8LtNVclEqPzcZjazSmeb 0cgq32lmvZ5MNJvSTcT0kjXZxij8/aqdtjCyFGdH4FPkBMtUJd1rQL4yLsteuebQlJpj8XH4FP7L of2bm6e8d777xPfxH+MyG6cX4l/gO9BRpEej9+oBIF8U2/FYLcKAB+txJdITDnaQWC4NGgN+1CLw CrYCdbYqFIyAIc92ql0g10AQKuRdajcjzIDCIiCLXZSyS0vL9h0dO3lgBfDQ0ca7IjWe6dfCc6/A 7WQ+WQjS3U/zNJAGjtTgGnhkGBGv0AAXePiGuymSdMbUr1BBTdeAQtQI3SwJOa4gubh9717mCoBf gvAGaD+HsjQ3oc2tTDRyJ+K3wvmt/JMJMWI8k2jWgaMU/zCNApEKoBGHxu9HXM8Xu+0VpL3nCy1o r3iQw4Tbwu3kCLccYTsNlmG4Ts+dRuQ00G47PJ7fsxLuXKme7VITdNgg9I/GVickMBp14CKMt7fE az3Cd7/YKd5O7PmatwgdwBNpeOIuQnWDpvcGeMEeMBpdAKinGf1pRfNQBtBZkIEeQU6DAbYGegwV APGPwuYo9If2yJfQMpfe6SzcSaR3+go4iVW+1zyKItJbqvQIUg0GuqXHem958Z5tYtCj+oE1d5Og 8j9g1jrha4WvueeUdj0vbiAblY3mN02CTlLcpNo2ynG1Z4hvgm2qY6rnGt8CaYEyw3aDY4Gn3reC 3CQuV1aaN4gPSZvVN90nyEfiR8pnZm9vc5fotFC4uFCHkU4FvdKSblmCqHIxwdEg0mDgWgJv3JVg fOD5WGO0K9lMHGtEMVRO/zB86+psqpUimtMKAsBwzqZS9LKogGiSOHHB+1uX71565fz3n/xgxX37 t69atX37rauujpH3MY8ve2nannjPiXg8/vsdD72CH48/+MMZPBfP/37eesorJ4GA54F2erRTC3JU 1y3gbyP3kodl/iUe65AoEE4nYAPBb+lZ6/W0TwhTLGjvOcUkHCrfahZGUD8jqIkRFEZZ81BypWjC 6OM1CBpoVyE1EoUCDgqaQASPcgBX4nUoIRyNURiXaOIPdhK2SpWrAluohgXYj4bCFlGUSkAOi8j5 tiven/DgXwuW8rdcvir95eFvTaN9qwRelqBvAfxGkpd0FtXottnEiUbKShYLq3yv6VQVagG7EKAs 6qIXBAL0bMBvgjMBA215oJ0c1AxE73IF01ULmC/pgAcFHxyl26OooIu2tIpujwykzEt6H2iwWgl7 oKYzW0jqOac0xWojEwN2eozeezfcmoqKopCJLoqQbBT/t6dRfqbPo09jD9NKBwuDxYPCq+JB6Q35 Tb80wlBnmGBaYJhpWmldabvTesj6pfdL3xmv4VXlFRvxqX41TQ2o4v/0nEESML8MpQ6o5Q3oVVkU 3/J77X6/V/Z7AS1kr58zBtR28syeMRZsacfuvbQHiA2HGRODfonrfRhtyuv4IFmDgkjF5ZrBsreK TCOLyG2EJwdIJkrH9+5KMDvgyrkohRdmfFZ1dcc6LVZKWdiklH8Ca1FKAspRDMcW19VlOUKRMqB4 SplTGGaaHjgB/nnpQhlxZT39yA/bHr7l9sfwftvP775/7qrnDz81NbBjxxWVMzpuPfLl7AW/eazZ duzTb3fUvnDomY3TBzC8ndTzFe8EXoniuiTpFI9boxRw+xGmzBo1wA7ODeuNZoM5oNfnOgJ+PpDr F3KNYaPB7QElGFQp+welCKUjvTxSQPHnaAH9IGtFVRUoki6gYNfr6uvWCvVIdCD9UgrmCEansdq4 3shXWyZblvu4a5w3qPPtM53LjCvs643N9jt9zxr1QpBjnEPnNHgJw3MxJYwGHTiIaajaiEvaDAYH 7z5AnkEeMlfLhlYK0Eyjdcm04KIgCbopLwebpCURhk4RjCJqhECLz75Cz0Ra8t3tuHy35318AJeD KunQlIt41a8d378rBVmMjhS1zkZjCeTq7qTsCbqSUjRBUBBWICHIK26ss5U5KWox0kllvdUUFSkZ JbpF4YzIpLb0BxbctvOp1UWj7FZlSfv6+fM22dtC375881sLZs+8vSV++qPXevAd7oc3tN6+6kn7 E+Tm1TNuX7s2uPeNObtnTnusf+B393TEf/qK+knICziggu2lh+GJaKXWWsNcwyOG7YY3DcIobpTx tzxnBT5HBpGTBL3CScgAAv8Wx9s5jueMiBiMvMQdJAeRDE7XVk2PeB4uQW/p+XYy+xVB0Gtp6cX6 FBrqE8qJVb5nWkrfjss0o6RlhIulplCJ1GImlKEUo70YEZUECUfoj+lvoNK5j/6G7DW1401srL8D BGRgeJZCTKX6lcqwEBy4c5WWigpmL27oH+VBbMxmMww4+Nf7kRH0vrUCYOcDTSmq4DLyKzg+La2S 3qIOyAHXaHaDplQYmsZWGLRIhSHDD2V+BUPcOnADS3CRpcgRtnAWTDZ3ryWP/+b119viJXjas9y+ C1c/G38SBPuB7gXAelT/h4TnAGcnJWRnP8LQPyPtEPab9AGHw2+lcKGYeT7gN5owktygM5hVwCpM ziiuUTmhnARs1H0EZIOKRq6V4a+ZbUd6V6Q1p222PW/7veEjw2c+WWdzm/K8nK5QKFQOAJZxIB+q Te+w2mxvmcx2k81uMhtBSDQbbYhm2goGp8msOXCyUa+Yefw+FSBANi1Im2eZpi5Sb1PvVXkVxMTN xMSNkVt1E3dKTNwtQeshXILM+AFgqvLdpr3/m7ikXyouFwUmRu1KkBLW0ZgFvtTv3SD3jwpARcTA j+EebgSL6xLBAWmxhRwhDiQGOewSWAORib9zPHzD7W07Nk3elLP9HvJp9ytj1t7XgeWld5/9Yzdu UpvvOvLUI7vHVDnJP16KL58aP/fuG/ftPsViCKgGaOcA3EtDeXhMEvnSzTgdT8Mc9uUENPDcjKCw fEJGwG7UBzDKUqkqY3acGnCplIYuhnsuZse5kkbX0Q+Oqn9I0RK8uyMxSsv8BR48VNIcQz1Dg1Os E4ILuJnSTHm+dWZwqbzMv05e7/9I/sBpkYJ0kLMTUiFODDPQo7UQOyHRE9nBcDBET1hoK8caCbTT h9+fRkkJwKdLtRms2nLNivZmLVEZKcFbUUFOoRdnXqF2idrST09pF8AVmrPKNc21yHWbi3c56TmX kz7O1U4y90QTphrIYlcvGZOox9AO+pikGRUginh1WAK/hRpookQBzkrVVDgDWdQyCnfY3oeo3Pk9 7n4jFky6YuL15IpDc9q6bzq+9i/xzsfvPL3j8+6yMfeMXvzMU7esfIEfb5pfWFN4+fd/nlEf/9d7 zV234pF4Fd7+2rbDFz6PvVDX/sRDO3cyfTYdMM8pPI+MqEEzHTFiHv6JzOsAz6gkFhLM6wzGJRxH 6KCMYaqaI16zvET3dzQGqD+NcFVQLMK3gRHpMSU5mQZpGitrznaNVs9Rq4x6CFSLV1gqEiobGJb6 MiLiRClcarWWTef2bop3jSw17+du/+ed/C87Nj0Qt8bPt3+2A3+L33iMxkHGAw96gAddKIwKCUpw YZsB+QL9KU6CPUYm9u9vDQVEISdgNQZ0Bspu1BvYx7yJqJn6mpQRzSkDilbYSbObSzmiXOoqrpeB uUyHgV7uYHd0MAZ2XPQaLnVJqOXVRV3zpGfyCmuImGqImGhIJ/NQzCkcTz6fHoPKBS2DHqSPpb90 MEhzsJ5e7F/qYfAsXJBsQOpLZaisxIlznSOcIyJfGb4pFHSFeDVajVfxS+VGZbFhmXGl6y7UjDfx 6+U1ylrDeuPdrncsr9usGSAru/1BLy2CwQJa5AcjVIACuUEDCriRAZqxtT/uM9JLXtVhXTuZo6nR JWYtCNJjxsismom5Hd+3b6B7SSs40XB+d+YSR69r49AcxNEyoNe1ScR5GOx1JfsWY51LhsvKk39g 4TWixro6fDGA02sPIDhis/eRl77Cg+c33PDVqx3fLli44e74uU8/jZ+77/r1C+auu3P2nI2DRrSM X7Ntx+23Pc/5ch+av/XEya2zH8ztd2TjoR6Ecce9r+EJc9feMW3GhrUXempaxjzXdPsL21I+LeXJ AODiy0l6K+mgBrIsoATOMYJSbcDwyU1dnRxKUbeFkdTCPB6L29IvquQEaJRjjIkzmexoLMbMmDSq 4F1gqm0yqDFNR+VINDaQwchANjBAbcp+KsXRz//Q61H0acRF/anlMQVqYVz8f3jqpc/61aMK+j5I Kx7kHeXUwtc6J4dnczc4F3rnhFd6Vwc2ee8KPOLc7j3k/db5VfBc0HaZ8wnnDic3KHemSLKp7g0D M7lDQTGYExhjmkYVrZ8+Er8/NgHKbbQR6QdwBVIAky2XqtaWfhSp2yhQW3p5yaJZiKUl+kZfm5Oy Uldf/ZkCXhRrxLG6pLa8nJQUZ1O8hRIBM1ktzHWO4FQUENinYYdz1fTxq8eW4tKDC/ddwNLr93bd svIfT710grz97NKbd29ftfpJPF5deeOo2z5pMLgnLcDyJyex+kj8b/Ef41/H97z8Klf86L4jj22i oEvQfgDe9XyExePLwZYQkCjpiFjJc5VY5PWkEmwbRKjn/KScjDI1UvwEn4DRIRH3o2E/+O4/evQo V3f06IXnjx6Fe08BPDcI7wM/ZqC1WgHQwkdWeVf5yPXeWT6ywDDdRKaAy0dKTUNNxOeRJR6p2RYL MubacQDRQHw4lBGqTNenV2ZkBCtDoQC6LnCj/jrX/Ez1uiA4dvPDk6ckYpDqOTrMlTT81c3CX+cq 2Th3Wpg5n4y6YhDSstIyOswXbXieiqyJSLQD+BMccA7IPFj+zE1LHnHv9/zr7Y8xmnJHbamXtB/F 8zKt82sGDY4+e/2geVtaHnYePfHtc/VPLR19df0N8QdpGAv6vCg+TvpQ+BANR5PRv7TJfEgNOkOh rBJjkanaNMI9NDQsc9iI4ZMmmFbmmpxZuTiiy0uL5JZ4SyuGZE1y16VdG5qUO2lE3aRZ7llZs3OX e1emLc5c517r3ZR2V2hDxGNSx5oQN56qQ705u1AZqxBFch4kV6EhaCQ52DZkEKdPpxbiIByMNkRJ 9ACuQdnk4L6CqzLNEpbayR2aWR17Ocq0bjVnFqoNYFQcwNuRjzzRVlWelwnX61CYPKHpgiW4xFM7 eVNCf9Z0dVNjMNZ1trsTXKUu8OW7YiCVncAHVbFOGOqkLqVOL4trU0BkASBXWRGX4N2yUmtJMckM Z/DEYbfyRcHMsiJR5MMZmZk0Cl5mRaGBPA2YMgsxO4LtSZkAYpkIf+cVT46r2zbv6R8XT36iImNP SyA3rWTS4nUvxncc/Ta++sMP8W9+wiK+vnZv0c/xF/7xRfzO+M9DJsxciV/D2s/4rsXT39n3SfVE uzHuvH1C+arGqzZM1xrna0+PvHbuJ2u24Kqt18Ye7Z6+yezLvmwsNt77PM54+bP4nG9/ij+xvfXW eSduW/zlA7/77Ozn2IyDb7+54+34F395Ky/bg0fd+dCQtW/P3rj5ipY/Af17uhES6sBvk5AJz9mH TWaVOUg/tiUrPzPwJVSb1zFFzJSqwLYFaqE6R56rq1c3ci3qm8LrYod6RlVkoQ5PImPVuUqr+k/D P43/NOl4A2/kTZyi1wk8D361LEqSAeqyaJAwQnTWyMyiWkHJYIdThOPoMQc9xgV5gx1+pQsIghwQ ObGdNGg6JBu+0Qgm5ABWQMkomtUQRLMk7pqx/DH+JM+18Jhvx1hTxho6pJMGrsWADXRfNUvHJHKb 1CQR6Tfmjz5OIIQHvvDvBu7welRAPndVpRe4hQloF43NRsFn2NDfzcrEPBQ4hOqRI6YjRzYIiRLk dmSrMn5ka2DclNo23szJ0oGeM3RqjAJnHV7cGEvE2sK4CIe5EGcLcZFsUeJI0buk9vMXux998lP8 j4eHZfiLhAO/DMOH4kPJFLx5/01330X9CA5tBnz6BmhlYZ6EbT/igSrDaRSW54eFJ4Vnh5fo1urE ed5lQoNuiXKHcIciZjt1nDs7L+BM0+ls1kBeXm4u8qcFYOTSAwELkt0R0UANNxE8aq2IKjvRShWd KNKxF2V6d5FRW7RTThAnZEUMfvoLg55eZ6Cc4aBXGbz90gJBFrQMJiOW55j2ZJVktPKXNkbmREVM xC/1LGYZiw6e6u6NR8bY9Cndqek6mwxRJmNZ8AWgBNysrCiw0OlAnAiB0HhlkSXUJ8ZhImEcGpgI ZEXC4G4PTEAp1DeTyLa3l8yes+7eyU2vbYr/Bl+2pvzqkcNufyL+GV54XWTIlEETHtgU3yEcqNs/ 67rnirIPNc3ZVT+Au8binF0zYlHu+a2SoXzBsGtWDKB+wOyer4XloDfS0Pt7Z5D5aQQnTGTWv9Pa NFoLooHGGagBLU1rQmvTWtAjwovcs8b9XJvxDeNx1Jn2zzSLyZpmSUvj8sQcS54/mD7cOMk+2THJ M1dYkHaL9S7rI9zDpkf82/AzZJvlQ5MN2ZFXtatenk477M6pYCZPfk6FakaY99kCBs4X4HVqxHw1 itBZYG+6KxKUsWygrZE9gRlTE3OtAJUw0LBNeloJBQQUoPFxcLIWYxeDPBg4ayZoH5cUobqIIiJV 73zb4cviv/+yK/7xozvxkMN/xv0Gv1p0+Dfb/zZ14Vfrn/4rIQN+OP8avvG9L/HEXafezt96/1Px H+47GP+m+VDCN34C8GcK8LQZRu9LrSCYjofICf60qAEzkqHROpzOgoQ6xlY6PZvncLMjjPkYLHnT 09T/mvn+lWK+n1PMF/g18yXrsYtMN6BwyAqtlPNJsigLMi/zosftdRNR0YMk6EFlOO1Om5MTfZwr hK0m2Lhlfwg79ZYQgnGMRvPgbw2OUR51OV1OcFUJcGhWaGAy1gp+bOgJ/O8Xp9xat3TJ6JX3HV0X 34Ur7nt2QHXNgzeM3hF/RzjgSBt1ffzYkefj8e3TB+4oHVD9zXNf/SsvAL1+CrCB5iQr6AHNIQoB WZYkxPF0IPW6gIJkifJHmmotliZwVwf1QSPRe4287v+HwBoGX5tgoeSg1TCRjdWc7Yz+WlIHFEKv HaHk9yk+88ITXPTCh9xa4cCOeNVLceMOKkfgFPDroA86dLcWZX24F0yAVDegC48FSVAhxKv8F+3W lMS8V1IM4//RfP3gqX2a36f9nQl3m1q9v277Nu7zC1+S1u6xtN2DdnTPZnGAhSD/+0H+s7BN8/rs Pgepz8bXyTZs5TIzUcjqIlkoQJiABmkrMBZdARMHHqAO40h2VmaQ46Bn2fUsSNnJ+sJ0cDJaeYLR gOlgH/09WdyUjbPTIkE91jMnSO+JzLi2V5xr1Ni5ZI+g+TRs0utaV7L9RNy4gpo/wNJD+bDP7/V7 /JxoiKhZjkh6RM7iI+EstzEthJxmWwguttuCEuxlCFkh7FeAt+0W2AR0oRDK5GCDWEAmSqc0K6Op P8rtYF+VZFkuQRCnS+pPAELonDg1q4D/LdwosvDe+PGtn8S3tO3BYz/bgvH9kZ2h6/ctWnf4plD5 Bkzuu/XM5aTqJdx9avGS/fi6Tz7CS9rmtP+2sKGpZtzaMRu3HIn/3DS9DFuAHs8ApmQwWfiExmg7 NK/NUcxzAZ1+q/64nugFQhQZZDgoSSKNHzOtB+NNw19QY4E2kTrtbqb9MNN+sSYjNhIlmJyV69D0 cNP/ggHlJAP2wRxnUn6CRhw0jjXWGxuM/OA6dzTW2Dsdl8CgBB2jlSwYDfIEPj0DIgyKDpgSvmHY PnOY/HL4cLcoHOh+jkz5ZRjZ013D+PJV2KyBceDQO3up/BA6Ibin/DI2MbinqDhR5hcmypzcRBnO SpRpgUTp9iYmEvOManFQaBF2CsCtYLTdi7aiVsQXIA2NRSfRGSRYg3CwBXFCIvJOx8GdHJ/vUuPz fWp8zmlqwuJj4/MU/1FdHwAeMrV2dxOYdbG6xsWV3bHUoNCQPBXHIsurh6mJBH0s6/mam85sou2a OovMEZeSZeJG40aLqGMS16ZQgWvHXk3hA2adLqLXyxGFOva0ZUoqMK4kEIJVEqqbHtFYgFKJBW04 aNNsY231Nt6GI4hNPSVg8dsUWf+cxJWR1n2pnnSpscZEj6gVCULYFa2iHkcy1FxaAh1h4crI4J1S w4wR83MO1712+2tH8Vb3tlVDltzK/XjB0/7W/C8oLantl8dyWxZqBkyAmwUkB6mRS57XzBLh/msg P/cfyk/8D+X3VSyB4InBDjk2HybvwYD/cwfjq4cQEs3QFpV0puYiZKAn43fZZLQwrAJCQ0WgU785 tGaw0tOC2cDpECayTjEhWUf0isjooCaJ8Ms+RgQV0QmeZF9+TvXlQtslSQw0ZFjV0aEeP95BZ8Gi UWa3RFEqqSFdYrwlsi3HtjzbCmwrU4kP0xphSgLgj2Kr6aKXo2dbKeUEyXTI0tkEnoANQb212Mw2 goFD2AQqVgZdSztO78Yq7CYHySRkhbGapBmT2khMEYDdFtEAZPRsASgiJuSVic7EEr1hfwxdoz7t NkTMsp34ZH65Yb3hjzCUhhGGEWYul88y9jPVctfyy403mzYYZYUIcoWx1DSGjOSGSppcY7zSpH+I PMxtljbL27jnJdFKzCZToUDsgkBkg9FYKMhQlQ3XmK/BGrhVMl3VBTJsMqmUTvXWJiuxHiDbkBEP 2C0E5XY8QNMbdPqgZrhNwcoB6KQJK3CGtIMzpjMDK5obVKy2k0mvBIV6oUkAWCDb9lgo0Hlo7k+s 0g2cxvwtqHt7dzpj4H1VVbK0v9THCz4Z9cI2rGZOGBQgRxedrd8hQ8954MGPwKH9iPlaI1sNcC4H zlEN8PMuk54eTU7afbAvVGHqF2ITd/vKKkwDy1h1bz4cTU7ORevAW0ONMRrrAgHATldpGQ4B3OIw tjyEM/G1hU5PCZ6GhYPxSTvjtcKB8z/ed9XYR7kLvwzj3z5fwp86H2Sy8hjoonRq0+DVu6xKSm/I boOTxchPayFak8HhDUoyuL4ykThO1vGE6CSZ54KiKKTQU+hVUUJClkCpaF7G0LGggoPKWKVeaVCa FEGRwT5iSsoID/vvDCX+P/VUr6HUB5qjsSjTTI1nL9FMLJ4CHjHPaJQQQJoXdeoVg6VYDsIGeLhu QCE1EoAKbbI2rAK637FvWIWsDUxUB1ZIGR6WRbXPA9WBiSo9Gk7kVinhCslkh6+N7p/dZ4NqWqKa BlUHrf68y5GkYBT3ER4gYhGm+hJbHnuDIwfeuBAHkq3hbwNyNZ1vSvgiM8CO+1z4AJmQD72ljfWa sV21230un4/nVd6uuBQfv921z/S6iXO53D4STNMsY2xjXJq3VqjVTVYnWqbZprimuSd5J/vucj1M VE+A46wBReeI0KxMqjUoGaSUFpTorAUdfInqETr+UiraLVHChBj8eJvScJo5Qqko9oEPjz/lwSVc uFgKv2suyXACN86m0jgVdTeYFVamoqKByFJMwI1DM/BGXPo2HvZiW3zfq8fiB7b9Ead9/Bn2rfjm vj/FPyZv4YX48cPxZ/98Mr517x/xlP+J/yt+DBdj3x6s/Cb+ZcJ/47uBv43IjXZr/WZZFtjJSHWk /Vr1WjuvGAKAMsjlTtjv1ojMYgaymsTfpD0re4NeDP9et/H/1qz/T6/E01eZJeMIjbFEJKHXsE9Y UWCcMmcsAM4sCYUsUO/1w0ju/TU33F/3ffzN+EZ8y6EnYqMGrI3fKRwwWWftW3gw3t39Eoc33Tb1 DocxwTu1PfcI3wPvOFAOnqs9MC2yJUI87jIHUfx8OrWw7en2sJgn5LuikcFCpWtQZJQwyjUiEhMm hmsji4RbuJXCJm6T8AB6hHsGvch9iD50fom+dH3p9vqFKMoTBgt8TLjfvTnyYYTPcuZFip0VkRHu Ef7q9OrwyMgkudYy0THFPyVtUvrk4OSMecJsx4LILZF7/PdEPnP/OeJR3NgBGLfbV4FoukK5r4J3 2915wiCBJ5wzh5NyIm6ngMQQZ/MKhO4gITMQMHNEzgxIOm/E5qa0sKV415aygWyUiSk1bCnepRUt i9LFdjXxBvOa8kheKAIIpTB7WmH8q3hyf82/NWd74zxdVSwQkfRTXBXIUqS+qb4ZS0Ym0GKKzI2L s2jWcbbYx7egXA5HS5PsbaG8XhbJ5n/asLjiicef/sMb8UM7W3H1m5Tlb+z+atvCF4HTP43/Ffv+ PHfqtbMej0U3VNxybQeeeuJTPPPAa/FnT+yNn7y7IPYYrtiN9b+JfxyHi+N/yh7sAZo/Cdi+A3jf jTLwBS1kVUzYWuqfkj5bXpjO61gqnsy2EttmUnOUDhlLjKMVQ6qipCrW9p6/7rF6i6E8sycju9hC 99Oyi9VkaU6WcP6TPWmRxHm4Xk2W9Lw2AipZpqv9VwfHK1P9C/2LdTebVpjX6TeaHzRuN7ebT5u+ Nqtg5wQtZrvFYraYDTqrj4S8Tr1opbl0glunc7q8noCLqhCW+ulyoVAGk2I38IFJDkRMj4mppFMx JaDMacpg7pPIgoWxYGZDZlMml5nh/m8lW/w/6qHw4G3/4bAnYc/T6e5KrhRISngUzlVWFLCct0TK m9CbXdvnDyW9DU0va+YKszrIYh1E1QVuZLaCCbSO11NhAb1kha9J81eoGXb4psO3V9HU9Qk7upwu W5jrTwBEwgxQ2Pxp6EnSfOSdlW+9X5MzcVTP2cMTb5ycHxr5F/zkus2jH3w6XigcGPPHFY99lJaV OXpZvBEPWLupXJG6l3FFZSuGz2U5pFN7vub/LryPColDy57BzeCXcEt5Piu7hKvwD+FGSKPSqtOH Zg7LHs/VSVPTJufcaTOFaQiBjndmqpKVqkRSlexUJcxIkbg4UclKVSKpSjb12obRWo4xkkkyueys UnNxeGhWdcGU4KTwxKwblPnGBabZ9lnuFcpK40rzanVZ5pKs9Vyzcqex2Xy3ui7zjqz7jZvNmx2B pI2eH4pYfRGvLpILbhXK9Vr5gQMiaBYIlzF/he9OH/FlOY35gewsnCU4BQosifmHQL4uEHByTNNF ASNiibAGLWIsJ66gK/HxaflZmSajIoT8aQGfLIk8R0SclZkBx0Qh4Mv3apTt7gXt0+VE+SxIw6wr FQfxWFyPG3ALFsGBbNVs+fSR9NHQ4qt1EZSLc6niNpnIxFzaNCP9Xa53IPQJR6zUbKOnrCkmt/ZO c1gnUFnwDEgGbWI1ncxb7GIR74uhWBU8Xzp7FT1LewRsTOcdaLS7jnqRjRe5GLDQVhYgRQOTccTM bDa9zxL+kvFah93l5F2MSSleRqa+Ypz2x9WLXhg/durg+A3j5s259cffPv3v9cIB847trU9WlONP a5tWrj//+Bvxfz6MP1ZvvHvylUuGVs8Ju6ZHy56etei1mfPeWWO66541144pKlqQM3jv8mXHliz9 JpFXXwg2wQE2p3SnZhRIAIYcsSWuunayZE8wMTPzihjEpICmNmC8FyejKqc1hQGEnESHH1NO619T MHEhBQvxhPNE7yjve7iv/0rXOKndnbGvVLa2oCq18KqERvCILZ7GN8d9gnHHjl/+mWjvk2D30XiR HX2q6SPmWr5WflPmnZQVnGA/F/OD5WH81fJy83PCabNkQMTSTg62iTp7hKRsc9JrmxM1Gaw7pfmZ cxkLOnHQOdZJ6p0NziYn5zSywF3KFdAHkwmNCUjUp7hF3wuJej7pUCYgUd8LifqYg5rmFyExGqPB v2TQIWEHMjsnimK4yJK0/1jUgQUgLHz94Znx8x/8Kf5Lw+HhO1Z/tE84cGHX5/ELT9+Djd9wYy7s fnXv9YfZCgakA103jOZ34suTGWxWASOZ2XV6JOhkAROh4POj6udHLUVFMOpVbHrVp2UWCDgP5XBZ +gJDoaHecKd8p67F0GE4Y1CChrEGwhNFJsnkDx02gBsNt6yqYnNs8Gu9TheUBbssCwiYhAh2QgQd POqboB780lkynkVkFm7KqRgr4ya5RYZ9jDUj0XIqphF8L9lCCKFHLEFhrEAKwRdtETqEM4IA/ujG PUr9toQ/2kjz4enXrSbWZHg9Xe7Euozk1B+d+Uv4nHbwK3cjM1DiH7t1VkwLcMvBNEok2VD3Mwcu K2XuJ2KLFZk5ThPWQrgo4U0WYXJF9x/fw6v7p2fk402vdx8Gj+Tjpoabb+ZzfxlGx9yDkLSc2hf4 My2SiyKWXGvEXYFKLRXWUvcINNwywjrcXYsmW2qtk93qQ/JD5uRAakUq9nqijmKh2DBUGGoY6Zgg TDBc65gpzDQscCwVlhpucZgFB41bWGUQNsLoWFXFqOZiCEoHP8DxgkBECQZfD5yoM5rMZoPdZrU6 nC63G8zJyj0CcgdpabBaaKlNcYDriQS68BDZMUZuQZYDDrfd4XBbDTpdwGGFqtViMJuDqsWuqhar ziC7HYLZooJcQZMEzq2azTqdLBNok9tqtViQ7HW5vOoVOjwOBZEBtg74akjA4/YF6dSWx9OO79qV MA5iXk9Nt9fd3e31dLtHV88a+lWvXZAKJlCbILmQMDWRW9M3tHBpAZJEV8YdgU3lkVSt7waIbQZi WyhPWPU0cSnBAVlwMO8iByTDFSY4ssegCVp5gikWx4AhbAmGsFmhsBXhMKbTwhg/Eb/ljZOZ3nI9 dn373piwP/+r38dvPBh/O1ty2eNvgqxWPfjA3zO5L7q98e/+eVcb9zI4s7FNwVnDzz+NUjI7AvjH RvZquaCTPNipkFxrrq0cl3Hlcrmu3DjIVGIts+mttqA1VGylGxMg1x4ojclSlyxlimg3QIWnV3F0 cxO+SSERPlfKUfJMEWspP0gepNA7XiVP4GPyVGWKaYJ1Dp7Fz5cXKPNMs6zL+JUytQxust5kW883 S836B/h2+RXr6/yb8sf8J/Knpo+sX/On5dOmr6z9RLbixWABVeykW0WmWxC2n/fQSpLTFQNy2FW3 3iJS6/O0ZqI1VUTECLhECAtOUCqDkkwwdQz4WafD9IULHCgbm9lkNGJVNVqsNpsCY0aMCmew6RUs qsSm09tsQaQDBNRxxGgMGji7wcABJnEcITYjKHwkFziwA/gzaNAMxNCOp70S1LfoO/Scvh23752W hJ92TS+2aepY9ZjKqXCRpg8ij91xOEThJzr6LOXamPtLT1esKwYVxrixSzh3g3AJk9L8Zfgzmylf VspH+hYJvjxSx8zbhJ/bG0pkZq1CgymeCkxNWrevwkpnkH0VtkTB07ROX4Wc4aug63R3+2lwrENL 91fYwPzl4Gs0OV2VNqvTdZkMfkIlx0NNoQGa/uCwZVgrFENa6DKM0kKVip7WCK0ZbC44ZnPBMVoj ULvEEo/iPnWwvwEui/BFrEwJhY6UxQ1fY/348IAhOPv97m4SPRO/Nz00wBFvIRfI/8Q3LqsaOxmv 66658G+i5JeMDcQx9c+v7jnN+/nLUQ4qI/laP51Rl+cxevNyjXl5FcZSR5lvUN6IvJgxljffOC+v vrDZuD73Eeej3u1GR05qRiCbra2jtec8L+Ts8xzMOeI5lvOe4/MceagTB6gFaKEK2mq9mCJTQqVm Iq2lu9Ld0X55xRV8Rb8R/FX9Jsl10dnyvOhywwbDm4Z/G/8dtZQVmzCvFmQWuwaG7O5puYtySa6/ wFRlute0xdRjEraYdpp+MHEmQ3Kl57eptZ9nNQddXWVieZEmkeZNmkx+ztVOXtjnfsDu90uIXuRl plN1tn6gn1Nyp6vTkcisqqxQJrVjkq7Cdwk7JpOntkcmnY+k2buZ1EOlfc+kUx0KfVwme1BmyibL bCfXaqZsja51CUYKIzsjQgX1eKkNDC7ER/tYZUAFC0sGwsWFFR0VZGsFrnDRtl1B7+jKcmcUZL4q HhNJulglEtHE3EuWMiy6mV/JUolFFv4RTczHZDOg4oDyPkvZwICOqqDKWSZ5r3FU2R398ktqE3VG U8toUtc3JtyH1HIaxJxEti4ANSayw6g5XcY+JcXZiWS8ywmzr50Outw5HOFoVl4i5xEu4ipn7p+/ 89DwJVeVLDgxBxdVb7xtRVqr+8bjd258Yayqc2Uc8ruuP7Jo6sCF8+Y+FUm7Y+KwF9eNXjPabjJ6 M7P0N+ZfVtfobrxrpDb96v43nzm/7rJy/HmOX82pKbiq/toxl93EYk7rgadpbJmuCW3SHsWCwZwp lAjVglCV3ppO0tMz/EX+K/0N6S3p4iBbpbPSO8o5yhuTY8Zac8x5nXe+fINxrvlG543ejvRPDSdc Jzx/tX3n+s7zt7RT6T3pnqBQYC6wFwpVZk0YZR4rzBZOpP3E/6IaVIeJFwny+UEe9Q6/SXFnHlew qmhKvdKk8IlsDYVxqeJOTuycS8U5zqRm1xLLTBW6zIAFhCgXFFCKKkuxpQjxiag0M/KLuCxCOjB4 YltxKz6D+XRchcdgDlM3gLItpmnfaZTBMGMWzIxwbKXMghmzUL3RRnmMXeqkj8ZuNhnP0pqwJzC8 7BJTmvLFYjqDDkfACbt4MBmeqmK5R4kEwsbFqDEUBrsaPK4AcagonJHNgcN1MXs6//m2xbuu39mo xX/83aEFpHjifctfenbZ8peEA90/3Tvm3reWxH+If/Q43vzqxLuOvn389aNMf4/tOc11AWZ58ZSk zV1sus2MzQqmU64NiEO81a9Ibj+vYJNDkmn/JdZ/iWUpSyrtv8S4/OgHryd8yiOxgfRLTfPhOgNO 9w+xDXGNt4131dvqXY+SR7lHjM+oz3gNstGjn0/mcfOFZYYGY5PxOcNe3T79XoPBaVhv+BvhTBnT zIvMt5k5MwaY0VYUsnngemhWC9qKTqEzYH6YzQq62EY/ND3TJDOMyvBB/zKVaDrYiJim6VESaYw+ VzGqeBlVRvgdmccknC5VSUQysQi5nl4kMYiVBviKjyQ9P6BLAgBii5Pv8mALpMrruhafjXYtTmVA WCoK1Fgn/DMPGihXh12JBSPJ5M+Ut8xeXFC5K+2Hl0/E/7X4mzt3/Dl9p+e2KRtfeGbt/HvwOtcr x3Aa1r+EyZqdT/oW3PD79z86fDuVyWFAs5OJHD08UXtGT3hjlrHYONQolNhL/JPJBP019vH+OWSm MEs3w17v70j/QPjQ9rnnS9uX9h9cf/d8yWTPmZ4e9VKBHeml0iv1J5nG/s5BpMQ4klQbh9lH+Cfr JxnnGL8Uv3b+gs+aVOzgTIpqBplUJAsCoeQUdxHNYTdnqepxC1YtmqXe0mQB4aQ8kRBRi5XKjoUp LiqsFpFykIWJrIU5tHTELSY64pbUfKaFup9XsgT8pdbMV6Vj0kmpR+IpicZInBRgLMewWgokWJGR jakmiWkgyRMoHts3c6ixpqu7r9ixVwFUdjL3lX4vShqdkwuVXJKtC1KH+y5aKJ915LYPl83/4I76 zQV7uoMvLVv+7LZbbn5y/RObzj+9BXPN464gpl+GEes7b732+ol3jlCajQQcDYCcOYBm4zVXOvI7 yEQuJsR0E5VZ3AJhkW6WIjsSb1ZgA9CpXUNraX62gsr6qfCL/ZyXH2Ad5Bngv8Ja473CP8461XON f7p1oXe6/2bxZsc5cs6tIic2G12usU4aCeCcfnOLulUlqsr7/HoJHSAvUI5N4VkHSAOMuwrS8YAN pMelgRn2ZxYaMKYWPRpT+QfMQtNl5xW3GrHRm07TNrIixbTUrqCqNh2nO4vUTEnLzCtOUSrYh1J+ RqmEgPkZjViOCqVUX1SMRWu6O0erjdHoucbeEANNz0guEqjsbqxMZtknEzKpFl2cErHExJNdCrHo Aw6xlVsid92Bft/v/yb+A7b/+UNswhdO63evm7Gp+wQZZyifdOeq7XiS6+k2nA5wb8A58S/i/1aD Ow/MxQ+sHzL3OYaTNiBik/A+cmGjFrDrsNlT4Cn0aJ4Gz6OGx4zbjbLXmGNs9XR4eA8dkRxvenGa bOQMZr8eO0jUbuM5Eem32LG9x6bxriweceR+zCZQ9wwoL2YTqVF/enELwh6NCopHM4KgIDuLVeWw WFUGFR3ULxml+jEZzLYng9nfMtXDkhrYYv32nl/Yujj0tNtzCB9AIXQO6xF4Auf6GMExOr96Fkz/ LrWrK0YDWZVsLXeFJZG4ZVctok4SZbCTVJ3Vhyyi2YfBjM5bswZHQVIWF1nCJUV0BQEICgAbxTUH XW26e8sWm/eO5aOm+soHXjP02DHukU2NC4qHTbY+rh9Wf/2mC7NBJq6Mj+O+BZmga3MWafWKItj7 KVn2UUq1XdSledL6KRF7v3CFUmq/WhlmnyTVKnOVX/Q/OUz9w/2yLw9fnj0qu6Xf1n5Saag0t6rf MGVYqDp3QmhC7jxpRmhGbn2/pn4nsk+Hvg//kG1xOUVHO9nVluO3SUyXqEFUyDRJE+pAxxEYr2S1 NlDw+8366gy/Qe90FGUV6bPc7uMurLo0V72rycX3gyEnE/sxYHMxYHP1ApuLARtdaMiOfpsANnoV XXiYBDYXNQyuZmsRl5pxFspIz3zVfMx80txj5tPNVeYxoOqYzJi9lLbmDLb0jcX4EstmzQzdzJ5o v6UhCnDg1l0EuLNd6q8wrrvzHF2h2plcZNOZCNU1glpy0fRQZkZmJ9bWUKRzlaSShfqu0Zq9Uxk4 ZOnqjW4TXt762Zkb37370MrnZn229X++ffi51au27Vh587Za77isgTOnlLXehSs/fwjjTQ81XZj/ 87GbX+Ty3u149Z3fv/57KkMbEOJo/qgdT9+PnMD4DlcxWwfPjOwsvoSr5g4YeXZokMtT7JLBNbdz AkZmvyDZFb0hS6cVlRb36HCHDjuZlnFqLGE3h23tlAQ66l5YWOous+90Xnqdjs4rMJLo7JQkOqpi 2FJUmuzL9s/tY2k5o1lQ1lVcWtzqPOMkDc6tzlZnj5N3EntWIuFBhTacoe9nCQLnnEI8myNOTjr9 ormYlPKpxLw+aQ+/JGxCRJhYEmZ2jnYMH9tnLpm9I4PlPkTP9nVcEy8VqUyYhDS0xKTTJJqkLJNo 8GGjDHKJqH+7BoFQJ5L3EovxLWELI6PosGxou7Vj+csj25YtGHt3JZiFP94fe+ax7mnkyQ23jL9n dfdBkMmN9N3PLJ9PQke163SltAdjdC26rbpWXYfupO6MTkK6dF2Drkm3JXnolK5Hp0/XgZUl8YTT idytGImCyOtFKUtA/BZ+K9/Kd/CneLGDP8MTxAf547DH8wl7mUzke8eNZ+PG6+lTeYZsfArZ+FQ8 nqdCpKdjyI+Wfz16i9n7vOhI9X2TV2xxY5QtyIJR2djW1sb//dix8w4+cv4EXZ/yVHwcHsT6bEUf atW8kCUM5ouE9YLgkgVB4nnCCzaEjQrh7AbeIigS7aEiSn6LuQUQ3eUCqTRm6fUtCk5XqpQxCkfT zLQy2qNk2hlzFhTmWSoB5p0YaKcUmfklTLYVj82+IzS8r1QzKaYZtqNVGq1pRFU1bNrampy2TgQX i4o2qHIid9wkq+aIrOp9WGeSfCjBEfS1REUOnHhTA43M07VE69viczNK08tK24queHAE/8277/77 lodNI+7np57feqRmJpVX4AXuZ5rjSqZrPjFhXYmTxCk6zmz8p3BO5HSpZR2JiWV9qqJLVVj2EZuY nsjdpCdWMWhjUb4ze6zZNOp3pg1Kq8AOsDDgGW0tHBF5XuDFMt1wIIWYr6/V38Qt05/g/iZKz4k4 LEakLLlCLNdVGccY6/g6sVaq063mVwgP614X3+M/EjvFb6R/if+WHVa9XuA4noiipNPJsKOT5SxJ tEuSyPF8lqC3C4JeDwzL0wkB+qZxGSQW6fl2bNZ0As9iLBky3QsFmX+gJtJlWsAEUrIQyQJ/EeEq NAYkh+ZyDmCyzyiOEqmnjJORlSEAcygQc06Qx2D8S2j47L60ZqRmM3SN59gMXfTifDMYqK4KOk9A 3w4BpTvxajUgu1zJsW0yVmkcqcPpurUc0bmNNCEKvI/EmyQ0va5fWoVOTkurFGkQLq1CpLkaQVbs CiXfF8Hy0RpRMoYn9nTsDrHEqd1OWnyxW60QEwXbM7Bil5LKZ8OJ4J9m/ZzHst0JT7PbK9mGTt/v dtMff7fLV5GcUKxLxEDo1CLDq0QITgIJxS98E5+PX/0i/uRtwoELh3BrfHn3TJK+Mn4t5cs7YFPG 5PVv+wQGUCyJuKw8kUxcXJIoCwckysQ7Azu0LFA3ZiFd2CKcFPgxsDkjcOlCg9Ak9Ag8oLmecAmA p3diQO8Ay2YLwh3gaJK+aP/zRbRP64P2CVon7DE5aYylphB7elKTiknsQqP5S7GLghcNICUSkDHb o390ZO5oY6nICR0qRsBmCuM3aJrh2VTW6NnUe7Q+0WoUY3EW38l36v7i+jIofCicCxKXHAzr3L6g juPCAb/ooCaFhMWw16Pqj2fhlqytWSQLcMyU1WLBFp75bCwZx8KCdcxns7NVyezNSrSjFsI8NwZj Fhams6RypSypvFNLO45pBndWiw/72O18vbfzsdv5aO6uhd7Ox7Skj7nePipLTDn7DPTGvlT8z0fv 50SkKJyFjyNMowAkHVH545j8pf2H/DHERc6kBr6QspHPanamihOkMCVEMjOrHd+859cInIjRdHf2 Cdv0CfjBTjeLpDeC/w/GM6geJsQWV9/1DyaD3RaxGyw+bDU6Uoo66bzQt7OwaXQXWyvP1DWzo/sq 7icHPjd/+YPpt771xAt7wlMvb/htW+3MUWsG8ZEHRk+7vvbAzn3d2eTxG6YNeuCZ7gfJ7ptvHvvI fd2fpmyur4BfnHi1ZhM40Ua2qe3q37ivbWe4czaRp5BbCQyzQsUPqcfdp9w9bj4o2012pxVsLiw6 jXqjyWDKdDM7y81sLoVZWwqztpRea0thQqBksCvoCDNrS2HWFuz/O0FQRZ+MyJ3TGBwqzKBTMPwr o91U6LzU8nKfcZMG91Z3q7vDzbs5UuRwMtk812axJFPN/1eDS/8rg8vSx+Dik5LYoVl/bcCNdrHl 6NGLYdzKs8wIu+RoIsmeJaOCDu61wpyiRaeX9ZKeE9WIRTT5sFlvTRKZLk9ppCjMqJyM5fYh8Yan ln1e/+RYVd+Wt+CqJc/zkQd3VjfUDFzdvYSsv3HhFfe/051cwza05zSfDXQ0Ig9esM/hTibBnWZi Rt+SoS2hNQ87YZX0HsNw8Sp5klgnzxHnyXKxOsg6yFnirlZHWkc6q91Tham6a9SYNea8xr1QWKib qS60LnTOdN+EHTpRMF7LTRAm6K813MDNEmbpbzDoXX5esgBo2DN9zPvxMUaQel8YJ7GATjIYmArA skoy0zTx6oxkNiqrdGi2zKziQgkjSZWCEicNOAkoQY+PoOEEqJsykcFEHV+2WhSxeCPyMwqzMEJS bhkCIbaMAmlwSwoIBA3w0rBC8mWUCdqpjdHYuVif5KzeLFYa86GKSzdeGK+7Xrhex1PtRC+xsZe8 oOQrX/q6RUOfufMPn2HnLX+/62S8a//uDet371m3YTex4ex7lsf/0n3077fjADa+8/Y77/7h7bcY DTfE5/EhoKEVBfD12j0GNV+9TB2p8lXB1iBJD+YawmkDHQPTrkxrCLYE5UGuQb6rXVf76uRrDVNd U33z5QWGeepC1wJfR/B9++fuz73vBzrtnYFTwZ6gM8xH1aijhB+kDuOvVqeoXyp/T4urisXEOf00 hC46/SYFmTyZx/VY1Wv6en2Tng8yIga1ZCrIV5rCskPcqdSQlFHXu9ogEU7XU24LszSRpdhWRIqs WQj975HzVMBc7RMwVy8JmJ/7dcCcTWkBTLKAefrwMje+JGKeCpj/OlyeSOes6Bstt6WA1emws1dG ZFu4PvTb8Myg++duPD5/2clbptzb3/Lc8ptffH7pkl3xecLvmseN29Tz0NPx83eNGtR9nnvm6JG3 P3z7rY8pDa+Kz+NOAQ1V5Mel2j0KiZI892AykqwwiFWOKs9IT0tga0AothX7qgJDbUN9423jfTNs M3z1gabAB+KH1q/EbwzfutVckmGIOipIiWEEGWaYQuaRTw2fuf/m/Mbzle8CMWPeaPf6Fckk2v08 EM5lKkI0ymrGqlkz15ubzHyABSMCjHpmFoww9wYjzCwYYWbBCDNTpiyc4KRjbU5k1YqJy6sYfiy1 /GeUNZPJMotDSCwOITkTxm8iapcWuDQC8b9EWLvPVv4nYVAjtiSj4aXJkMMlsdV+eQ9O/N3/09iz gEdVXD0z9+597eM+drOvbLKbTXYTspHEEAgLkdzEYIBUAgSRhISHCApiAQOileLS4vuB4gts/2LR +mzLkvAIYAuKTxSlLWKLVWnFVv1LUX+0RZvkPzP3bsRqv6+7e+acO3PuY8+de2bOzJlzB04t/d2a 55dv6S/6+TXdj269euXDA4uINHYyHo7FhwZ++OidX5zP/eLQoQMvHjn6ouVfdQPcnBfgvujoZXNs pRdrPC7ma/jz+TZ+Ib+CF2RdkiXZ7dVlN+Ik7GQPBVLksrskLMVjXuwlcf0/2/dDPb5/mvpZzY3A lNHX+hWWiS+c1dWfbDQ/9w0T/4TWdfoqugqUCiedC/+GtJdv8rCFJ11X0VW8VgW2xtVEaC5u2DJu Uf2s2eMaG8fO9hXyyZ8unzDmsdLm+rlX9R+x5FA/+AG3DeRQxQXM6/i4Lz5GniQ3lcyIL4ivlu+U 15U86n2q4lnOLQfCwUBVS8XRgCOfXESIVo2VYKfUKXcqnc5OV6d7sbRYXqwsdi52LXZvT24vVanL X8mwUSUdSrvz0uSlZSuKV5RkSu5RfuzaUPZAxX1VjyhPuB4ufaSsN/l80l+W65HGc0RxjijJEWWW lWjzUKI4R5TkiAI65W8Upjuk0oRL4cOxZB7vHF4QpkN48VAFm2cI1YdaQ3NCW0OvhwQ1FA0tDb0b 4qOh9SES+hXcnTyoGWzU2/RRdo0uPdLwYTD4sIbZ+r1en7/GGg336DUYD+8sWFJACiJ5Im9NSLMB ir/kBiH+YnrpLeYjw53RMA6XhExvsKaa7l7Jxm2DVkqfmBCLixuK0T1DMbpXiBmQITbyHeojs3rE knLYdUckfbgcl9Oz0D3Kc17P5blnFYiPWByq8jA7VVFpec3c6v3VpL46U02q6Qh+CQpa/V5W6WKW lEG9U4JeQIxFy6IXEStRmRJW2eWpMVtLfGHGmO5gq9Ps4cb4uznzNnSuPUwPD3rOiZ7GNQJ01WR7 IjyVWn5WLIWUNSfGwqkuZxPh1KahzqYUDa0RDli9KLP0nMJih68iqWuG5tU4Ie6O5SO5TMzHjnMg KfTBZpGnOB/Fi90uaZiSj8tKZUVI8fkoqhXQ/pa1MpglzNGjPLV27Vp0lkqi40BdQ0H4SpOlwwkN NP4NJ1b40vUabCS0vke95brV14xM3PPCptaG0eV3t33/Vx161tW9aPViv78yf92+B2YseuH7r/8B nxe54qoFTecVBxPVE9dObr62LJqacN1lwWmd02qLIwVepWREw+rOjs0X/5w+pyWDn5JyxyYUoOuH FbokNlnDQtY3AJEJYYRdbgVzyK/JKVWB5ptzqlocxbHbSLjwoCiNl8fPFZeJGfEukUfQf3pIzIr7 xcOiwJb+2GuATrNaJFInVzZxa9llNmGvCvqC1Q7aM6PtPx3isTtoVt9S3EMWoyAetW3hvxmrLLB2 f512gmr5k9QDlmp5fcQI7WXLBT4RsCbR6AyBXstiUDKfUKKFv1N3yZKKdet6d+zwpsoKf7pZG7dg C5l/OxaXDNxxe/89F1aEmZ0Puuw4fS8abt2NwnT2CSx4EvP66RKEj80Rhq8m5cUlktfvwl6/E9S5 DmJCI/yJYICaFWFmswSYtRIw2PD8kOtJgCnwwJCdEvDZA/X2qHCAGZ4Baqe4qTwGA3h/AAcmh9m4 ADVRwh+HybLwQ+FseDDMh10JeajpoJGiY/Jh+bjMy7mmQx5qOuxRaYWNRdPjsxZDZjaKzAaF5cmh rw0N0MHfbxojdf1s9rK+LhdmCB6iMK953Kqb+lPSABJgkPCufOSWdGsosLx8reWmZc9vlibZcGDg q6XCXP3qN2Y/3Ko5tzv1706deufY7T/ePuHK1pHdZEN/7x3nNk9tW38zSdNhU7g/YTqeD/dHwR/Z vgMBh4QUScDCkLtuCVtVWJk622uXOe3uGunAKK6nFarh3XpaBoOzRqIJAV3XCxjbWKGDGnJhUQ0q g4T1PuV4ogb5IYGtY+aasuE1KAaJ6hqGyuSkkkYjlQmoWZmBZ5B2aaa8EC8ki6RF8jVoFV5FrpWu kVcpN+GbyI3cLeLN0q3y/6CN8t3Kz9EW5Vdol7hNeRk9rxxDbyh/Q+8pX6LTSgX8HSWI/EoZSiq1 SisyFdlhGv4aB1SWmlxMauqrLNBOBa1UKnOeRkyLUlnQPNappVJhucThcDmpg9DbKZANwKHUoRSq HHJqrlVESUrIik+WFcQRkrC8XR2KAt0W5roqiIrMIeyodGFXXDJN03rvBc7fYToyDuIAypRjxMRx 50e/pfXpZDjU39XfFQ6ePNFlByobGmHU019f/Er9/23/pLOWsjDfYuZK6h2B8S8Hlvz6RCIaTP1t 98B3+WT/usuWTr+a3MyqB11KiJBjF9QPgy/IreE2aA+VaSDLLUywbY0jLMwyz7zzKaXHXFbB/u0e a4IAmldK6SbbVnQOIxf0ibCggjzcLhZKyKVjwiu8rtgjVZay02nAxUPa0UPaEbac2/ZAZv8v5xSY D0+hD5fzwxQySZ+l36lzeswK+WuHLuVzhE5VjxwtqtEiBdYYtrkrWlLDCy7ZK+TLIcPBI15wyk6P ZGjIy/nEiJTvLABbNiGWSylPDRopjpHGepq4ZsEUL5RanOerzfokY5Y6zbhCvFS6zLhW+J64Qtot 7FF3Gp8JX8plTr0MlblLPWVqqVHpG41qjVXSjdJG7gHXY/hx8rjzUdcOtFPY43mJPyr8Qf6A/0D9 q3Fa+EKOONkKKRdLNcFy3GPNOksNu+LmKx6VN5AuiVJCVBMeas55RM6NXQl33+BRs5ZqKjfUv3Jm s7mxzysoTj2ppPTp/DSlU1+ir9Zv1RVd4aE20tth3Zh/d/auTJ2utJaZaCfo1+oBwC/f9HHMCVx0 yIoiga2iaLoOOr6l14EM6LdMNBcqqid2QBelmKgbRsoh+hwO0QP3OeH2+NxujwRmT0qRfLA79Qy3 nxVEsGjwkqq7PG52eQbochq1hj48hkrXjSq+zzU3puExMm7O3YcfM5VYq4KXKtcrROkjF5lyq46X 6tfrdEHGRaZTc+C5bMyYg8frsR34c+/nC1m3KHTh6a6uIPRt4Ecfs67gt3uF28+dztL/wimcvjeF AqUptGSjbTO3u2OuGHl68Dj0a48jz+Dh7ahKjRlQR4eicLa3ZGvaWNSEw9tEGlcRMoraWrIjmNuS NHh8mxizcg17hTtdinZ4J3QH4digrw73iFX0iD1oNNljnWno4EP7Bdh++uDxXiXGx9Bo2+PcXth2 ZKeRRhUGWw66zUuH/du/Gulm6+LY6neqUphG8QaYazpXyuGWgb17nqjnRzyxe/PI83ZuHdi+94lh b4KK+dEJ/SD5bv/GVw6RhV8eI6t3/Ot1pmtUaIs+AV2j4T/abVGeip0CT2SBCG6okyrrl6uVKVYt WQSq/F2qgdV4iM1pmFNC6Q71fv5+aZPnQXW/Y7+wX3xFlVXTnw5zXjnPHdZG4jHOtfhOp1RpXMy3 i+3OmZ4H8EZlo3MX6XO95DzoeVU7xr0h/8b9lva+YhhfeY8buhp0aznvcUqpzHtcUYjwTe/xhYLA Wf7jgsw8yFVVow7kqurWhrzHNUVQiapoL6AXZKIlhvzHX3Bjd+JsF3JBYy7kSquBjYnuNa64os4T 5DWmAq3DLlOYImRYuLvzTU+MW0PirSDLifrq5+wI/6zBgPZCe187ffIb3uLDU112fe2yw11TZ3Hm If6clQJiL/6R6uxqsd0TLEgzR25nQdoVD6Q5ALrdU5TW2HqlvDSOF6VlM5Ie8tRmQ6hsxoi6awdo 81NL54q4UqzidQOb/vTw8EhFovfNgbvxbW8fGzPwISnDA2eaqxpHfDng6n8NT2of6IL/VTQwlfs7 1JEw/oddRwoUn8o5uUhINQSn4DUNNeY0XTG7roQqU+G3w8FD4ZBGETPWWcOR36tGsEr/xJWRdJlv hrpV4Uy3CTckVlZVo9FEdMmG3x00Sp2lrlL3KNco90jPJt1ZZpR5J/jbjXZve94iY5F3Ud61wtXu a/Xv+b6Xd4P7Vv1243bvLb6NyuPOp7W9+h7fR8pffZ+5+7UzvsFIYa5G+b3OSD6vNqnrVE4NDV2+ NZhgDC2tqVVVlwbaEnoPIZ/XmzAUH2yoLlCHCacCxrDipW7kToEeAEW0CKmM7IuQSB+p36GCLExf H5luOusN0yBzjH0GMfpw404Vx9H4fIUWMWmZMVeVq9XFTXENsrUJjb2VKsiG1G/Pj60G1QjC66dx D6ES0cgbQe30iRB9L8jJcFA7ySgUpOZDrkZJZ09w0iplrzpoyXpA3wRB3+xFrsEPkHPwA3y2tvEN vrOzNq3Ea9MeeMp25KV1e/FsO+01I+jHQPXxllo+L7VsKYzdjaGvnCiOX+8bW1E3IaAnHc6BK599 OxWPpt7bPrCkoaRq9Yyagcue0MpK8q9QC/iy/k0r166+mlzx5UtbG9vbqO4pA91zBOqVB2813UYf eVkiBq42AnSm+zVTBgKPK2Tz3s+ak4AYRsrkSi2N08pEfAG5QJoot2qdeDqZLnXIU7QleD6ZLy2W r8MrpOvk2/AN0i3yGXyaBoNN4mFSSk5LP5PexCJ9WnZpeTUEFKxM19sXgzlNxsgKkRQlgQk0gATT MJhkniMFf1GZ50bW20tYe57yKKQPq9uhOXQIe8ks0KIiHb5iA/dx90MejDymZ64n4/nY42D+/yW0 yLMCKWsw3opwK1qKBhGH2IJ7FFK1FUVUbdDxQHsmu58SJ1LMv4yFoE3Vae+Dofg+c7a0u5ua5zk7 qM7yLmSvHtkxDCclOjRjSU+isoStZ3dRKVJRWqHAlrfjLnbvJVAfKhWCjT7YlZ+WJX/+ebR71hNI M+NL8aeJDyDs/0qxjBiJhWK6aBOLo0YU5ZWRR7pnDrRyl/Y/s/Taxfh/N3CSsGFV/+zr5B+xGGz4 MLmcewU5UXQ34nCb6ZGFV6kfGUErXRfn3vxVefLcKm8u7jVYWVfed/mi++5bdPl95LVF9967CGga Z/MUiNtg74US4Xh1piE4COaVOto74XlOUeqgIx/qEeugmxHaxdWhV1x/+IyeoZ9+bT2snQxUW6+M KrLh1CH8ziH89muH2Afa7H/hg/xSuLscKjRVPJK+vYqtC2Yvhzr77VUcSIHnu/HBu++m49xt3P+R Dsfv4NoC6Pdm5+bQ1hA5JZ7yknfFd73kdfF1L9kn7vOSreJWL9ksbvaS9eJ6L1kjrvGSL6UvfWSJ tMRHOqQOH3FJLh/xeSURbGon4tQzHu4M8bgJdtW5UR2N6z/FrPQuFa8X14uciL2jfXUet6sOumdm IFzjWYnF0VIdwaiO49YTTEJB+xV+Vuw5rf8Eew0No1A9DcZxUjtpSceKxww/RCM7QIf/quXLl+Pl 9gd34bxiFvw4IAhi0Vk09j0TK59VUVvD4XtzFP/cb352Y92UYRcEZl38FcXmdZq5D8lkx8tMVm+Z k5msPpY+9hEsYR85Lh73ksPiYS/ZL+73kqyY9ZIt4hYv2SBu8JIfiD/wkmXiMi9ZIC3wkTapzZaV 6nJyyPeUl0rH5QaheUBcWHpKpBlVGERIUB3GHrXOBRIrdQfGuVxuKjD3SkKgvoDQShFdubg4aL+U ro5OmtQxYZ3QGM2iQtNXouXw18U1JKnly0FylheRT7TemTbiLPriZ6KpWRWjRnK/zxH8P0FEY6cO a/bPafuKYu80Q5zr0CbvX56do9Z9JuVL7I3Jv27LCjlMvbFYtGjozzB+y5ZE4riByeh8DX2xdWCE NmaoJPdpFZD1rmuALHkTzXbMQDPxTaiDPIlWM0ijJwE3AN5D+fhudBHAuwB1ADMAwnbehQDzANro NvDuBtwhRdFSx4zBfjju/Y4X0UKAnwC9hX8PPS6k0ZWw/Qjw7uMRqgX++4Un0UbI+zGUzQe+nwCe Cds/BboT9qmyaVm8A4Uohn0mAdwI+08BfAFACxzPC7gR4Cb8IroZvzi4BcoBox/C8W+i+QBNNp4A /+8GKK+H/Upg+4dAh+E8AmAVoAigjBxCV4LUT/Hdg/8C/jaAZpDfKPjejo7jo/gLMobcSQ6Sv3Mu romP83c5HhXuFZukuOyRL1V+6XzJ+blrmXuZ5wLVULPadG2nbujHjNnGYW+T99a8WN7n/rsCvqAn 1BR+MPxp/rrIuILZhSdik4vGxYPxT4t/W7I6sSJ5QXJ/6XfKNg7zlw9PGanTFUfP2TI8WflF1e3n dlYvqdkycvuo47VHar8YPXz0pNHH01ezu90KST2ajxzwxGmoEs2Ae+hWaAtEWA0x7DohIA9Ck5ra WifPSDVctWjekgunMw74DJbSGHff8mllsY959lZUBZ5jF9gTGtLhmDQGbh7yw5MdZO+lz0cRVIDi qBiBQY+SqBRa/2GoHKVQBToHrqoKnYuq0QgEZj5ItBaNRmk0Bo1FTWg8ugAkPQFNRJNQC/oOuhBN hvNOQVPRNNSGpqOL4B9djGaidtSBZqFO1IVuQk+hHrQd7dg2XW2IcwF0CmAQgENRSCsBWgHmAKwH 2AwgINXOWQpwPcA+gI9ZickFejaMMPsA3cZQ7+Il1WxznrXZ2cU2ey9ut/CFUy3cNNFiG2OxnVtj ZQ9vtHBphYWNRHWGYsVdvb/Bz/nRYQCClkGKyXNIBYM8ih7i8lAWgHCCnWNyRm9JsnrzPo5HmCMc Rpei6OB+Dve49eoGhQySU3AfouTv5KRVQk72evTqzQ2TyJ/RVoB9ABz5M3z/RP6EridgqiIN0nqA zQD7AF4HOAUgkOPwfRe+75B3kEreRpUA9QBzADYD7AM4BSCStyHVyB9ptWIppesBCPkjpBp5C/7W W5Cq5BhQx8gxuLTf9dSmq3czIlVpE9GETQTybcLwV/eR3/acGRbtI+/1xlLRhxqqyBHQW0doTYVU A4gBTAGYC7AMQADqKFBHUQbgLoCHALIAoDsh1QBi5CDAqwBHURWACTAFQCKHe+A0feT1nmRjtMFP XiMvQpWOkkPkJYZfJS8w/Ap5nuGXARcCPkhe6CmMogYnlCPYRwOsAa6Ecgd5prfEiA426GQfiCcK aSVAPUArwByA9QAC2UfiPZdGDTjIXnQQdE+U9KAPGX4UbZGQuThqJs+HOhajSXLMeUBBsjm2OUnM 5P2bYJMmyTs3AEWT5LrbgaJJ8ntrgaJJcsnVQNEkeelioGiS7JgDFE2SrdOBgqSP/GRXSWm0tvUK HGtQySqQ0iqQ0iqQ0irEk1X0i87w9Np+1FNeDhJ70EwNK49m9uDM0zgzDWe24MwCnFmDM2txpg5n ZuNMCmciOFOIMybO7MWjQRQZbG7/2mbaDOLMQZz5Bc5040wSZxI4U4IzMVxr9pGinokjGBrPUG8D fa4AnzeuWoVrLAKJFkG1LoLHfh+krwMMsi0TmGJxizlUSHG8t7ze2h4+pnppwwRyAHY8ALfhAHoX gIcbdACq0QE4yAE4gAppPcAcgP0ApwAGAQTgjsOFr2epCmklQD3AHIDrAU4BCOxyTgEQtNS+xK3s wirti26lW+QAfOPwLSJFZoEW0VLaBG492M+FuLVwsJDUIr8f1K+hS3ofdu/8h/uf/3AjuUGGRmg9 qNooucvG63vOFET78Mae5N5oQx5+ABXyUOtwGiVxAvBo1M22R6KIRHENipCnAFf3RGbAbmpPsiK6 B3voXjujZyInoh+CpQvkB5G90TdjfTzuib4BOU/tjB6J3BJ9ubJPgpynk30Y0J4YY90dGR39xUHG uhYKHuyJrqFoZ/T7keboFRFWsMAqmN0NW6YanZbsiE6A4zVFLoma3XDMndH6yOxoncU1ku6zM1oF l5CyyHK42GERdtLiQnbAi2r78OVmhXi/OFNsFUeJ1WKFWCRGxQIxX/RJhqRJHsklKZIkCRIvEQlJ PrqAIkXbRJ+gsYF4nqY8ozVCU6vJRNBDJdAcZb1cC2lpa8Qt2f3zUcslseznbcV9WJnakXUUN+Ks 0YJapjdmR6da+sTBadnaVEtWnDJr5jaM72yH3Cy5uQ+j6TP78CDNuiE/a5xPl0Zi/YY78ikuu+GO 9nYU9F9dH6w3xunpC5q+JZlrp2dNNwS/Rhdk729pm5l9sqA9W02JwYL2luw9bTGwAPGn+OPxTbvx JxS1z9zNjcOfjp9G87lxTe3tLX14BuODDvEnwAc15hPGJxWiGOVDManQ4nvQ4kvA/sBXQhHwyTJK ML6ELDM+HlO+bd0l45u2lZQwnkAMdTOe7kDsbJ6DCeBJJBiPP4MOMp6D/gzlyY5jLJEIsBRGGAsO owhjieAwY5nxFUulzXLLEMst7Ewc/oonYvG4j+d43MeBJ/XffhY0gmHdO7Z9fuf4BcXj5xaPXwAw N3vb1ZcHs5lLYrFt89tpQSzLJedeMv9yiuctyLYXL2jKzi9uim0b2/ktxZ20eGxx0zbUOX76zG2d 5oKmnrHm2PHF85rae5un1NR+7Vy3DJ2rZsq3HGwKPVgNPVdz7bcU19LiZnquWnquWnquZrOZnQux Oj5l5jYJNbaf32nhXuJUoL7OzS9qb/Rry8axyju2KLgmfw90SB5HzlR71lXcmHUD0KJzGs5poEXw TNEiD2SrdlFwzdii/D34cbtIg2y9uBGlVqzsXomC4xc1Wb9u+EDWipVU4Faa6v5PHygbnzXnNXWv QKglW97Wkq2f2jFzmyhC7lz6l7JjcnlO5/i+wf1W5nDIHEMzOW6IkebV0TxZthm/ef9Xnh3DI0P2 9mKzEK9A3e1ctrBlOgFVML0D/mtnx8w90F2izUN3O/zBbpzC3bljsMu2J9cQ/b85WLHSpmw5rLCx tRfs0p0Tx9CHSgn9Pyz/d2AKZW5kc3RyZWFtCmVuZG9iago5MiAwIG9iagoyNTI2NQplbmRvYmoK OTMgMCBvYmoKPDwgL1R5cGUgL0ZvbnREZXNjcmlwdG9yIC9Bc2NlbnQgOTA1IC9DYXBIZWlnaHQg MCAvRGVzY2VudCAtMjEyIC9GbGFncwozMiAvRm9udEJCb3ggWyAtNjY1IC0zMjUgMjAyOSAxMDA2 IF0gL0ZvbnROYW1lIC9KRFNPTlYrQXJpYWxNVCAvSXRhbGljQW5nbGUKMCAvU3RlbVYgMCAvTGVh ZGluZyAzMyAvTWF4V2lkdGggMjAwMCAvRm9udEZpbGUyIDkxIDAgUiA+PgplbmRvYmoKOTQgMCBv YmoKWyAyNzggMjc4IDM1NSA3NTAgNzUwIDc1MCA2NjcgMTkxIDMzMyAzMzMgNzUwIDc1MCAyNzgg MzMzIDI3OCAyNzggNTU2CjU1NiA1NTYgNTU2IDU1NiA1NTYgNTU2IDU1NiA1NTYgNTU2IDI3OCA3 NTAgNzUwIDc1MCA3NTAgNTU2IDEwMTUgNjY3CjY2NyA3MjIgNzIyIDY2NyA2MTEgNzc4IDcyMiAy NzggNTAwIDc1MCA1NTYgODMzIDcyMiA3NzggNjY3IDc3OCA3MjIKNjY3IDYxMSA3MjIgNjY3IDk0 NCA2NjcgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA1NTYgNTU2IDUwMAo1NTYgNTU2 IDI3OCA1NTYgNTU2IDIyMiAyMjIgNTAwIDIyMiA4MzMgNTU2IDU1NiA1NTYgNTU2IDMzMyA1MDAg Mjc4CjU1NiA1MDAgNzIyIDUwMCA1MDAgNTAwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3 NTAgNzUwIDc1MCA3NTAKNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1 MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MAo3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUw IDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDM1MCA3NTAgNzUwCjc1MCA3NTAgNzUwIDc1MCA3NTAg NzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAKNzUwIDc1MCA3 NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNzUwIDEw MDAKNzUwIDc1MCA3NTAgNzUwIDc1MCA3NTAgNTU2IDc1MCAzMzMgMzMzIF0KZW5kb2JqCjcgMCBv YmoKPDwgL1R5cGUgL0ZvbnQgL1N1YnR5cGUgL1RydWVUeXBlIC9CYXNlRm9udCAvSkRTT05WK0Fy aWFsTVQgL0ZvbnREZXNjcmlwdG9yCjkzIDAgUiAvV2lkdGhzIDk0IDAgUiAvRmlyc3RDaGFyIDMy IC9MYXN0Q2hhciAyMTEgL0VuY29kaW5nIC9NYWNSb21hbkVuY29kaW5nCj4+CmVuZG9iago1IDAg b2JqCjw8IC9UeXBlIC9QYWdlcyAvUGFyZW50IDk1IDAgUiAvQ291bnQgNiAvS2lkcyBbIDQgMCBS IDEyIDAgUiAxNyAwIFIKMjIgMCBSIDI2IDAgUiAzMCAwIFIgXSA+PgplbmRvYmoKMzUgMCBvYmoK PDwgL1R5cGUgL1BhZ2VzIC9QYXJlbnQgOTUgMCBSIC9Db3VudCA2IC9LaWRzIFsgMzQgMCBSIDM5 IDAgUiA0MyAwIFIKNDcgMCBSIDUxIDAgUiA1NiAwIFIgXSA+PgplbmRvYmoKNjEgMCBvYmoKPDwg L1R5cGUgL1BhZ2VzIC9QYXJlbnQgOTUgMCBSIC9Db3VudCA1IC9LaWRzIFsgNjAgMCBSIDY1IDAg UiA2OSAwIFIKNzMgMCBSIDc3IDAgUiBdID4+CmVuZG9iago5NSAwIG9iago8PCAvVHlwZSAvUGFn ZXMgL0NvdW50IDE3IC9LaWRzIFsgNSAwIFIgMzUgMCBSIDYxIDAgUiBdID4+CmVuZG9iago5NiAw IG9iago8PCAvVHlwZSAvQ2F0YWxvZyAvUGFnZXMgOTUgMCBSID4+CmVuZG9iago5NyAwIG9iago8 PCAvQ3JlYXRpb25EYXRlIChEOjIwMDUwMzA4MTQzOTUwLTA1JzAwJykgL0NyZWF0b3IgKFBvd2Vy UG9pbnQpIC9Nb2REYXRlCihEOjIwMDUwMzA4MTQzOTUwLTA1JzAwJykgL1Byb2R1Y2VyIChNYWMg T1MgWCAxMC4zLjggUXVhcnR6IFBERkNvbnRleHQpCj4+CmVuZG9iago5OCAwIG9iagpbIDw0ZTli ZTMzNzgzNDRkNzMxMGVmN2IyN2VkZTJlYjk1MD4gPDRlOWJlMzM3ODM0NGQ3MzEwZWY3YjI3ZWRl MmViOTUwPgpdCmVuZG9iagp4cmVmCjAgOTkKMDAwMDAwMDAwMCA2NTUzNSBmIAowMDAwMDAxMjE5 IDAwMDAwIG4gCjAwMDAwMDAwMjIgMDAwMDAgbiAKMDAwMDAwMjI2OSAwMDAwMCBuIAowMDAwMDAx MjM5IDAwMDAwIG4gCjAwMDAxMjAxNTIgMDAwMDAgbiAKMDAwMDAwMTM0NSAwMDAwMCBuIAowMDAw MTE5OTgwIDAwMDAwIG4gCjAwMDAwMDIyNTAgMDAwMDAgbiAKMDAwMDAwMTQ0MyAwMDAwMCBuIAow MDAwMDEwODgzIDAwMDAwIG4gCjAwMDAwMDIzMDQgMDAwMDAgbiAKMDAwMDAxMDkwNCAwMDAwMCBu IAowMDAwMDExMDEzIDAwMDAwIG4gCjAwMDAwNzQ4MzYgMDAwMDAgbiAKMDAwMDAxMjg3NyAwMDAw MCBuIAowMDAwMDExMTI1IDAwMDAwIG4gCjAwMDAwMTI4OTggMDAwMDAgbiAKMDAwMDAxMzAwNyAw MDAwMCBuIAowMDAwMDkzNDU4IDAwMDAwIG4gCjAwMDAwMTU0MTMgMDAwMDAgbiAKMDAwMDAxMzEx OSAwMDAwMCBuIAowMDAwMDE1NDM0IDAwMDAwIG4gCjAwMDAwMTU1NDMgMDAwMDAgbiAKMDAwMDAx NjUzMiAwMDAwMCBuIAowMDAwMDE1NjQyIDAwMDAwIG4gCjAwMDAwMTY1NTIgMDAwMDAgbiAKMDAw MDAxNjY2MSAwMDAwMCBuIAowMDAwMDE3NjI4IDAwMDAwIG4gCjAwMDAwMTY3NjAgMDAwMDAgbiAK MDAwMDAxNzY0OCAwMDAwMCBuIAowMDAwMDE3NzU3IDAwMDAwIG4gCjAwMDAwMjA1OTEgMDAwMDAg biAKMDAwMDAxNzg1NiAwMDAwMCBuIAowMDAwMDIwNjEyIDAwMDAwIG4gCjAwMDAxMjAyNjEgMDAw MDAgbiAKMDAwMDAyMDcyMiAwMDAwMCBuIAowMDAwMDI0NjA0IDAwMDAwIG4gCjAwMDAwMjA4MjEg MDAwMDAgbiAKMDAwMDAyNDYyNSAwMDAwMCBuIAowMDAwMDI0NzM1IDAwMDAwIG4gCjAwMDAwMjY1 NzQgMDAwMDAgbiAKMDAwMDAyNDgzNCAwMDAwMCBuIAowMDAwMDI2NTk1IDAwMDAwIG4gCjAwMDAw MjY3MDUgMDAwMDAgbiAKMDAwMDAyODc4NCAwMDAwMCBuIAowMDAwMDI2ODA0IDAwMDAwIG4gCjAw MDAwMjg4MDUgMDAwMDAgbiAKMDAwMDAyODkxNSAwMDAwMCBuIAowMDAwMDMxMjc1IDAwMDAwIG4g CjAwMDAwMjkwMTQgMDAwMDAgbiAKMDAwMDAzMTI5NiAwMDAwMCBuIAowMDAwMDMxNDA2IDAwMDAw IG4gCjAwMDAwODMwNzYgMDAwMDAgbiAKMDAwMDAzMzM5OSAwMDAwMCBuIAowMDAwMDMxNTE4IDAw MDAwIG4gCjAwMDAwMzM0MjAgMDAwMDAgbiAKMDAwMDAzMzUzMCAwMDAwMCBuIAowMDAwMDM5MTc1 IDAwMDAwIG4gCjAwMDAwMzM2MjkgMDAwMDAgbiAKMDAwMDAzOTE5NiAwMDAwMCBuIAowMDAwMTIw MzcyIDAwMDAwIG4gCjAwMDAwMzkzMDYgMDAwMDAgbiAKMDAwMDA0MTg1NiAwMDAwMCBuIAowMDAw MDM5NDA1IDAwMDAwIG4gCjAwMDAwNDE4NzcgMDAwMDAgbiAKMDAwMDA0MTk4NyAwMDAwMCBuIAow MDAwMDQ1MjU3IDAwMDAwIG4gCjAwMDAwNDIwODYgMDAwMDAgbiAKMDAwMDA0NTI3OCAwMDAwMCBu IAowMDAwMDQ1Mzg4IDAwMDAwIG4gCjAwMDAwNDY3MzAgMDAwMDAgbiAKMDAwMDA0NTUwMCAwMDAw MCBuIAowMDAwMDQ2NzUxIDAwMDAwIG4gCjAwMDAwNDY4NjEgMDAwMDAgbiAKMDAwMDA0NzYxNSAw MDAwMCBuIAowMDAwMDQ2OTYwIDAwMDAwIG4gCjAwMDAwNDc2MzUgMDAwMDAgbiAKMDAwMDA0Nzc0 NSAwMDAwMCBuIAowMDAwMDQ3ODQ0IDAwMDAwIG4gCjAwMDAwNzM4NTQgMDAwMDAgbiAKMDAwMDA3 Mzg3NiAwMDAwMCBuIAowMDAwMDc0MTA4IDAwMDAwIG4gCjAwMDAwNzUwMTkgMDAwMDAgbiAKMDAw MDA4MjQ0NSAwMDAwMCBuIAowMDAwMDgyNDY2IDAwMDAwIG4gCjAwMDAwODI2OTYgMDAwMDAgbiAK MDAwMDA4MzI1NiAwMDAwMCBuIAowMDAwMDkyOTAwIDAwMDAwIG4gCjAwMDAwOTI5MjEgMDAwMDAg biAKMDAwMDA5MzEzNCAwMDAwMCBuIAowMDAwMDkzNjM4IDAwMDAwIG4gCjAwMDAxMTg5OTQgMDAw MDAgbiAKMDAwMDExOTAxNiAwMDAwMCBuIAowMDAwMTE5MjM4IDAwMDAwIG4gCjAwMDAxMjA0NzYg MDAwMDAgbiAKMDAwMDEyMDU1MSAwMDAwMCBuIAowMDAwMTIwNjAyIDAwMDAwIG4gCjAwMDAxMjA3 NjcgMDAwMDAgbiAKdHJhaWxlcgo8PCAvU2l6ZSA5OSAvUm9vdCA5NiAwIFIgL0luZm8gOTcgMCBS IC9JRCA5OCAwIFIgPj4Kc3RhcnR4cmVmCjEyMDg1NwolJUVPRgo= --Apple-Mail-1-1007160049 Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Disposition: inline Content-Transfer-Encoding: 7bit _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc --Apple-Mail-1-1007160049-- From speechsc-bounces@ietf.org Tue Mar 8 16:40:32 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA22774 for ; Tue, 8 Mar 2005 16:40:32 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D8mTr-0001qa-QE for speechsc-web-archive@ietf.org; Tue, 08 Mar 2005 16:43:07 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D8mQw-00020s-QO; Tue, 08 Mar 2005 16:40:06 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D8mQv-00020n-AV for speechsc@megatron.ietf.org; Tue, 08 Mar 2005 16:40:05 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA22757 for ; Tue, 8 Mar 2005 16:40:02 -0500 (EST) Received: from salvelinus.brooktrout.com ([204.176.205.6]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D8mTN-0001qD-86 for speechsc@ietf.org; Tue, 08 Mar 2005 16:42:38 -0500 Received: from nhmail2.needham.brooktrout.com (nhmail2.brooktrout.com [204.176.205.242]) by salvelinus.brooktrout.com (8.12.5/8.12.5) with ESMTP id j28LcfvE008739 for ; Tue, 8 Mar 2005 16:38:41 -0500 (EST) Received: by nhmail2.brooktrout.com with Internet Mail Service (5.5.2653.19) id ; Tue, 8 Mar 2005 16:34:44 -0500 Message-ID: From: Eric Burger To: "IETF SPEECHSC (E-mail)" Date: Tue, 8 Mar 2005 16:34:44 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" X-Spam-Score: 0.0 (/) X-Scan-Signature: 01485d64dfa90b45a74269b3ca9d5574 Subject: [Speechsc] Slides Posted X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 08e48e05374109708c00c6208b534009 Speechsc Slides are posted to the Supplemental Web Site: http://flyingfox.snowshore.com/i-d/speechsc/ _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Wed Mar 9 15:42:33 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA18692 for ; Wed, 9 Mar 2005 15:42:32 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D983U-0002Iq-CJ for speechsc-web-archive@ietf.org; Wed, 09 Mar 2005 15:45:20 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D980g-0003a8-Sb; Wed, 09 Mar 2005 15:42:26 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D980g-0003a3-BU for speechsc@megatron.ietf.org; Wed, 09 Mar 2005 15:42:26 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA18680 for ; Wed, 9 Mar 2005 15:42:24 -0500 (EST) Received: from mail97.messagelabs.com ([216.82.244.131]) by ietf-mx.ietf.org with smtp (Exim 4.33) id 1D983K-0002I5-Kf for Speechsc@ietf.org; Wed, 09 Mar 2005 15:45:11 -0500 X-VirusChecked: Checked X-Env-Sender: Skip.Cave@intervoice.com X-Msg-Ref: server-6.tower-97.messagelabs.com!1110400934!22630087!1 X-StarScan-Version: 5.4.11; banners=-,-,- X-Originating-IP: [205.143.75.32] Received: (qmail 5570 invoked from network); 9 Mar 2005 20:42:14 -0000 Received: from unknown (HELO SRV-EXVS01-DAL.intervoice.int) (205.143.75.32) by server-6.tower-97.messagelabs.com with SMTP; 9 Mar 2005 20:42:14 -0000 X-MimeOLE: Produced By Microsoft Exchange V6.5.6944.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Date: Wed, 9 Mar 2005 14:42:14 -0600 Message-ID: <6E80E3E8D788BA4DB7EEFC88FBE9B0130467C600@SRV-EXVS01-DAL.intervoice.int> Thread-Topic: Recording functionality Thread-Index: AcUk6HnPveWfKP3ARReLG50xryo9Aw== From: "Skip Cave" To: X-Spam-Score: 0.3 (/) X-Scan-Signature: 93df555cbdbcdae9621e5b95d44b301e Subject: [Speechsc] Recording functionality X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Content-Type: multipart/mixed; boundary="===============0219938838==" Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.4 (/) X-Scan-Signature: a743e34ab8eb08259de9a7307caed594 This is a multi-part message in MIME format. --===============0219938838== Content-class: urn:content-classes:message Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C524E8.79C61863" This is a multi-part message in MIME format. ------_=_NextPart_001_01C524E8.79C61863 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable All, =20 I have a question about the "Record" function in MRCP V2. Can it be used to record both sides of a phone conversation, while TTS, ASR and users are all interacting on other MRCP control channels?=20 =20 =20 Ellis K. 'Skip' Cave Chief Scientist, R&D Intervoice Inc. =20 ' 972.454.8800 (office) ' 214.460.4861 (cell) * skip.cave@intervoice.com =20 =20 ------_=_NextPart_001_01C524E8.79C61863 Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

All,

 

I have a question = about the "Record" function in MRCP V2. Can it be used to record both = sides of a phone conversation, while TTS, ASR and users are all interacting on = other MRCP control channels?

 

 

Ellis K. = 'Skip' Cave

Chief Scientist, R&D

Intervoice Inc.

 

' 972.454.8800 = (office)

' 214.460.4861 (cell)

* skip.cave@intervoice.com

 

------_=_NextPart_001_01C524E8.79C61863-- --===============0219938838== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline Content-Transfer-Encoding: 7bit _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc --===============0219938838==-- From speechsc-bounces@ietf.org Wed Mar 9 18:22:10 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA02608 for ; Wed, 9 Mar 2005 18:22:10 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9AY0-0005tD-Fi for speechsc-web-archive@ietf.org; Wed, 09 Mar 2005 18:25:00 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9AUX-0007uV-JL; Wed, 09 Mar 2005 18:21:25 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9AUW-0007uQ-70 for speechsc@megatron.ietf.org; Wed, 09 Mar 2005 18:21:24 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA02540 for ; Wed, 9 Mar 2005 18:21:21 -0500 (EST) Received: from e32.co.us.ibm.com ([32.97.110.130]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9AXC-0005rf-Mb for Speechsc@ietf.org; Wed, 09 Mar 2005 18:24:11 -0500 Received: from d03relay05.boulder.ibm.com (d03relay05.boulder.ibm.com [9.17.195.107]) by e32.co.us.ibm.com (8.12.10/8.12.9) with ESMTP id j29NLD5j654650 for ; Wed, 9 Mar 2005 18:21:13 -0500 Received: from d03av01.boulder.ibm.com (d03av01.boulder.ibm.com [9.17.195.167]) by d03relay05.boulder.ibm.com (8.12.10/NCO/VER6.6) with ESMTP id j29NLDZJ204654 for ; Wed, 9 Mar 2005 16:21:13 -0700 Received: from d03av01.boulder.ibm.com (loopback [127.0.0.1]) by d03av01.boulder.ibm.com (8.12.11/8.12.11) with ESMTP id j29NLDar027395 for ; Wed, 9 Mar 2005 16:21:13 -0700 Received: from d03nm120.boulder.ibm.com (d03nm120.boulder.ibm.com [9.17.195.146]) by d03av01.boulder.ibm.com (8.12.11/8.12.11) with ESMTP id j29NLCHo027388; Wed, 9 Mar 2005 16:21:12 -0700 In-Reply-To: <6E80E3E8D788BA4DB7EEFC88FBE9B0130467C600@SRV-EXVS01-DAL.intervoice.int> To: "Skip Cave" Subject: Re: [Speechsc] Recording functionality MIME-Version: 1.0 X-Mailer: Lotus Notes Build V70_M4_01112005 Beta 3NP January 11, 2005 From: Jeff Kusnitz Message-ID: Date: Wed, 9 Mar 2005 15:21:11 -0800 X-MIMETrack: Serialize by Router on D03NM120/03/M/IBM(Release 6.53HF294 | January 28, 2005) at 03/09/2005 16:21:12, Serialize complete at 03/09/2005 16:21:12 Content-Type: text/plain; charset="US-ASCII" X-Spam-Score: 0.0 (/) X-Scan-Signature: ea4ac80f790299f943f0a53be7e1a21a Cc: Speechsc@ietf.org X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: bb8f917bb6b8da28fc948aeffb74aa17 It seems to me we discussed that long ago. I don't think it's something that logically belongs in MRCP really. It seems like a more reasonable place for this would be upstream from the reco and tts resources, at the point where the input stream comes in from the network and the output stream is sent to the network. This has the added advantage of recording a caller's input even when not recognizing. Jeff speechsc-bounces@ietf.org wrote on 03/09/2005 12:42:14 PM: > All, > > I have a question about the "Record" function in MRCP V2. Can it be > used to record both sides of a phone conversation, while TTS, ASR > and users are all interacting on other MRCP control channels? > > > Ellis K. 'Skip' Cave > Chief Scientist, R&D > Intervoice Inc. > > ' 972.454.8800 (office) > ' 214.460.4861 (cell) > * skip.cave@intervoice.com > _______________________________________________ > Speechsc mailing list > Speechsc@ietf.org > https://www1.ietf.org/mailman/listinfo/speechsc _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Wed Mar 9 19:32:12 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA08529 for ; Wed, 9 Mar 2005 19:32:11 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9Bdm-0007MY-Nn for speechsc-web-archive@ietf.org; Wed, 09 Mar 2005 19:35:03 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9BZP-0000At-Ed; Wed, 09 Mar 2005 19:30:31 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9BZN-0000Ao-EB for speechsc@megatron.ietf.org; Wed, 09 Mar 2005 19:30:29 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA08422 for ; Wed, 9 Mar 2005 19:30:25 -0500 (EST) Received: from salvelinus.brooktrout.com ([204.176.205.6]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9Bc4-0007KX-AW for speechsc@ietf.org; Wed, 09 Mar 2005 19:33:16 -0500 Received: from nhmail2.needham.brooktrout.com (nhmail2.brooktrout.com [204.176.205.242]) by salvelinus.brooktrout.com (8.12.5/8.12.5) with ESMTP id j2A0Rm2F000458; Wed, 9 Mar 2005 19:27:48 -0500 (EST) Received: by nhmail2.brooktrout.com with Internet Mail Service (5.5.2653.19) id ; Wed, 9 Mar 2005 19:23:50 -0500 Message-ID: From: Eric Burger To: "IETF SPEECHSC (E-mail)" Date: Wed, 9 Mar 2005 19:23:49 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable X-Spam-Score: 0.0 (/) X-Scan-Signature: a87a9cdae4ac5d3fbeee75cd0026d632 Content-Transfer-Encoding: quoted-printable Cc: Skip Cave Subject: [Speechsc] RE: Message Control X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 6d95a152022472c7d6cdf886a0424dc6 Content-Transfer-Encoding: quoted-printable Copied to the list for everyone's enjoyment. > -----Original Message----- > From: Skip Cave [mailto:Skip.Cave@intervoice.com] > Sent: Wednesday, March 09, 2005 6:51 PM > To: Jeff Kusnitz > Cc: Jesus Bernat Vercher; Dave Burke; Eric Burger; David R Oran; > Paolo.Baggia@LOQUENDO.COM; Sarvi Shanmugham > Subject: RE: Message Control >=20 >=20 > Jeff, >=20 > Can the element be issued at any time during=20 > playback of the TTS or audio? For instance, the user is=20 > listening to a message, and he wants to make it louder. He=20 > presses the 3 key on his phone, and the browser issues a=20 > command to the speech engine. The speech=20 > server should immediately turn up the volume of the audio, in=20 > the middle of the playback. The remaining part of the message=20 > will be played at a louder volume than the beginning.=20 >=20 > This goes the same for speed, pitch, and skip-around=20 > navigation, all of which can be requested at anytime during=20 > playback of a prompt or message.=20 >=20 > Skip >=20 > -----Original Message----- > From: Jeff Kusnitz [mailto:jk@us.ibm.com]=20 > Sent: Wednesday, March 09, 2005 5:30 PM > To: Skip Cave > Cc: Jes=FAs Bernat Vercher; Dave Burke; Eric Burger; David R=20 > Oran; Paolo.Baggia@LOQUENDO.COM; Sarvi Shanmugham > Subject: Re: Message Control >=20 > Assuming the content is SSML, then the element can=20 > be used to=20 > control the output rate. I'd expect that this would only cover=20 > synthesized output, but the SSML spec doesn't say that=20 > (http://www.w3.org/TR/2004/REC-speech-synthesis-20040907/#edef > _prosody),=20 > it says: >=20 > The prosody element permits control of the pitch,=20 > speaking rate=20 > and volume of the speech output >=20 > So perhaps there's a synthesis engine out there that speeds=20 > up prerecorded=20 > audio as well (I've not looked to see). >=20 > Jumping around can be done using the Jump-Size header from=20 > the spec (page=20 > 35 of the current draft) >=20 > Jeff >=20 > "Skip Cave" wrote on 03/09/2005=20 > 12:54:07 PM: >=20 > > Another question: > >=20 > > There is a general industry requirement to modify the playback of=20 > > messages during playback. The requirement is to change the volume,=20 > > playback speed, or skip around in the message, all under speech or=20 > > DTMF commands from the user. This is a very common requirement for=20 > > voicemail applications, but is used in many other types of=20 > > applications as well.=20 > >=20 > > How does MRCP support this key requirement? I would expect that=20 > > there would be a set of MRCP V2 commands to invoke these functions=20 > > during the play of audio or TTS messages, but I don?t see=20 > them in the=20 > spec.=20 > >=20 > > Skip > >=20 > >=20 > > From: Skip Cave=20 > > Sent: Wednesday, March 09, 2005 2:47 PM > > To: 'Eric Burger'; 'Jes=FAs Bernat Vercher'; = 'Paolo.Baggia@LOQUENDO. > > COM'; 'Sarvi Shanmugham'; 'Dave Burke'; 'jk@us.ibm.com';=20 > 'David R Oran' > > Subject: Recording Functionality > >=20 > > All, > >=20 > > I have a question about the "Record" function in MRCP V2. Can it be = > > used to record both sides of a phone conversation, while TTS, ASR=20 > > and users are all interacting on other MRCP control channels?=20 > >=20 > >=20 > > Ellis K. 'Skip' Cave > > Chief Scientist, R&D > > Intervoice Inc. > >=20 > > ' 972.454.8800 (office) > > ' 214.460.4861 (cell) > > * skip.cave@intervoice.com > >=20 >=20 _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Thu Mar 10 12:29:54 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA09050 for ; Thu, 10 Mar 2005 12:29:54 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9RWn-0003fV-7P for speechsc-web-archive@ietf.org; Thu, 10 Mar 2005 12:32:53 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9RRq-0000ev-S9; Thu, 10 Mar 2005 12:27:46 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9RRp-0000ee-00 for speechsc@megatron.ietf.org; Thu, 10 Mar 2005 12:27:45 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA08845 for ; Thu, 10 Mar 2005 12:27:41 -0500 (EST) Received: from sj-iport-5.cisco.com ([171.68.10.87]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9RUe-0003cC-Ph for speechsc@ietf.org; Thu, 10 Mar 2005 12:30:41 -0500 Received: from sj-core-4.cisco.com (171.68.223.138) by sj-iport-5.cisco.com with ESMTP; 10 Mar 2005 09:28:10 -0800 X-BrightmailFiltered: true X-Brightmail-Tracker: AAAAAA== X-IronPort-AV: i="3.90,154,1107763200"; d="scan'208"; a="166477456:sNHT342915590" Received: from vtg-um-e2k1.sj21ad.cisco.com (vtg-um-e2k1.cisco.com [171.70.93.55]) by sj-core-4.cisco.com (8.12.10/8.12.6) with ESMTP id j2AHRWYO017863; Thu, 10 Mar 2005 09:27:33 -0800 (PST) X-MimeOLE: Produced By Microsoft Exchange V6.0.6487.1 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: [Speechsc] Advance Material for SI/SV Security Discussion atSpeechSC WG meeting Tuesday Date: Thu, 10 Mar 2005 09:27:27 -0800 Message-ID: <6677B3346233B94EBB11C060935101200A17553D@vtg-um-e2k1.sj21ad.cisco.com> Thread-Topic: [Speechsc] Advance Material for SI/SV Security Discussion atSpeechSC WG meeting Tuesday thread-index: AcUiSLNCKN61ozhmQmeq5IY1paFYFADSINfQ From: "Shanmugham, Saravanan" To: "David R Oran" , X-Spam-Score: 0.0 (/) X-Scan-Signature: a069a8e8835d39ce36e425c148267a7b Content-Transfer-Encoding: quoted-printable X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 76c7db407a166e4c39f35d8215d8dd32 Content-Transfer-Encoding: quoted-printable I agree with Dave's analysis here. Further more, what we are trying to achieve here is no more risky than what is already supported in the IETF protocol or achitecture. 1. If we are concerned about the storage of the voiceprint on a server that could theoretically be broken into, the question we should ask ourselves is, what is the risk it adds to the system(network+server), considering we already transmit and store voice in these systems. And these stored or transmitted voice streams, if compromised, could be used to generate the voice print for the user. The way I look at it, generating and storing a voice print is no more different than transmitting & storing the voice in a different code(though a highly custom codec, that cannot generate the audio back, but then so is the DSR codec). =20 2. The next question we should ask ourselves, is the value of the voiceprint. Does it really compare to other biometrics like a fingerprint or a retina scan in how its applied in verifying identity. In speech verification we use a challenge response method which would require the caller to mimic the voice on the voiceprint. In that sense the voice equivalent of a fingerprint or retina scan is a sufficient segment of the users voice, which we know is already allowed to be stored on servers attached to the network(voice mail)=20 3 If we are concerned about the security of the audio used for verification, it is no more or less risky than today, as speech, today, is carried over RTP/SRTP and that speech is used for various purposes, including storing as a voicemail, to be recognized for information such as passwords and PINs etc. The fact that we are applying a verification algorithm instead of a recognition or storage operation does not make it anymore of a risk. =20 Sarvi=20 -----Original Message----- From: speechsc-bounces@ietf.org [mailto:speechsc-bounces@ietf.org] On Behalf Of David R Oran Sent: Sunday, March 06, 2005 4:32 AM To: 'speechsc@ietf.org' Subject: [Speechsc] Advance Material for SI/SV Security Discussion atSpeechSC WG meeting Tuesday Here is some advance reading to frame the discussion. The IESG has a "discuss" on the SpeechSC requirements document over the issue of security and privacy concerns about providing speaker verification and identification services. This note contains some points I gleaned from reading the document cited by the IESG (and which is also already cited in the security considerations of section of the Requirements draft). I'd appreciate any thoughts you have on this to inform the discussion.=20 Please post as response to the list so we have a record. If any of you wish to spend the $22 to otain the report, that wold be very helpful. It's probably worth the money because there's a lot of good stuff in there (despite my disagreement with the conclusions related to SI and SV). See draft-ietf-speechsc-reqts-06.txt for the exact citation. Thanks, Dave (co-chair SPEECHSC and draft-ietf-speechsc-reqts-06.txt author) ------------------------------------------------ I thought it prudent to read the entire NRC report which Steve Bellovin used as a major justification for his "discuss". So, I swallowed my annoyance at having to spend the $22 to download the PDF (otherwise you can only read piecemeal and can't search). Here are what I believe are the relevant quotes from the report: "Recommendation: If biometrics are used to uniquely identify license holders and to prevent duplicate issuance, care must be taken to prevent exploitation of the resulting centralized database and any samples gathered. (6.3)" - Box Es.1 on P.21. "Biometrics has the obvious advantage of authenticating the human, not just the presented token or password. Common biometrics in use today verify fingerprints, retinas, irises, and faces, among other things. =20 Downsides to biometrics include the fact that not all people can use all systems, making a backup authentication method necessary (and consequently increasing vulnerability); the fact that revocation is not possible for current systems (the saying goes that most individuals "have only two thumbs"); and that remote enrollment of a biometric measure (sending one's fingerprint or iris scan over the Internet, for example) may defeat the purpose and is easily compromised. " followed by "Recommendation: Biometric technologies should not be used to authenticate users via remote authentication servers because of the potential for large-scale privacy and security compromises in the event of a successful attack (either internal or external) against such servers. The use of biometrics for local authentication-for example, to control access to a private key on a smart card-is a more appropriate type of use for biometrics. (5.2) " - N.B. This seems to be the key point of contention. - External attacks against a speaker verification system can be foiled by well-understood security means, and these are in fact employed by speechsc - TLS encryption of the control channel, SRTP encryption of the media channel, and authentication/authorization of all elements in the chain from the speaker to the server holding the voiceprints. - Internal attacks can be simply the stealing of the database or the compromise of the server system, including its keying material. There is the pragmatic issue of a server with voiceprints being like a /etc/passwd file, so it is of course prudent to store the data encrypted so it can't be stolen en masse. But speechsc is addressing a protocol standard, so it isn't clear why we need to say more than we do about that aspect, or more than is said about server security in the Radius specs. Server compromise can leak voiceprint information. At issue is whether it is the use of the voiceprints for authentication, or the mere storage of the voiceprints which causes the security concerns. For internal attacks, it appears to be the latter which is the privacy issue. - Speaker verification differs from other biometric authentication techniques because it is easier to make it resilient to replay or impersonation attacks. Unlike static biometrics like retina, face geometry, fingerprints, etc., speaker verification can (and is) done via challenge response protocols. The subject is asked to say a number of words or a phrase chosen by the verification system, and the result matched against the vocal characteristics captured in the voiceprint.=20 Possession of a stolen voiceprint does not by itself enable impersonation, as would a fingerprint or retina scan (c.f "The Sixth Day", "Mission Impossible") etc. - voiceprints of varying quality can be obtained by anything which can record your voice. A voiceprint used for identification/verification needs to be protected, but reality dictates that not all recording devices will be super-secure, hence requiring those holding voiceprints to be more secure than those holding speech recordings seems questionable. The "security first" tradeoff would say well, then all voice recording things need to be as secure as the servers holding voiceprints. Are we prepared to go that far - revisit all the specs like RTSP, SIP, XCON, etc. to ensure that they cannot be used to make recordings that can be turned offline into voiceprints? - an accurate voiceprint is in practice a biometric of the same confidence as a fingerprint, retina scan, etc., with the exception that it can't be used unless you can be induced to speak enough to allow a match (sort of like the old call trace scene in the movies where the bad/good guy hangs up before the listeners have enough time/info to trace the call). Superficially it is more subject to surreptitious acquisition and mishandling than other biometrics, but in reality I don't believe it is given how easy it is to lift fingerprints from identifying objects, get DNA samples from personal items, etc. - speaker identification can be done by anything which can hear you speak. If you speak in a non-secure session, there are obvious problems (including surreptitious recording, as noted above). If you speak in a secure session with authentication (by means other than speaker verification, rendering the identification issues moot) it seems there is an implicit agreement that your identity can be ascertained by the participants, by your voice or other means with or without the aid of computational services. If you speak in a secure session with explicit anonymity, there is a higher degree of expectation that if the context is one in which you are not identifiable by other means (e.g.=20 historical, correlation with prior sessions that were not anonymous, etc.) that no voiceprint exists, or if it does exist it is of limited utility (e.g. it can identify you as "the one who's been interrupting on the last three calls"). While the existence of speaker identification in this situation does limit privacy somewhat, it is not out of line with common expectations of privacy in live communication situations, or after the fact for that matter. Those who are truly concerned that indirect methods of identification (such as speaker identification) be thwarted are the people who do things like use voice distorting devices, rendering the identification and verification systems impotent. - Many other authentication methods also have privacy weaknesses of various sorts. As the report points out (cf. p.67): "Private secrets are items of information that are so intimately associated with an individual or with events in the (human) individual's life that no other person (or few others) would be expected to know about them. The use of private secrets for authentication causes several problems. People resist the use of private secrets for authentication on the grounds that they are private and should not have to be revealed to third parties (even to third parties who wish to authenticate us). Private secrets are rarely completely private.23 This leads to another problem: Any item of information that is used as a private secret to authenticate an individual will typically be shared with all the people and organizations that want to authenticate the individual (technical measures exist that could prevent sharing this, but they are not widely used). Each party who authenticates the individual therefore comes to know the information that is supposed to be a private secret, and thus the information becomes less private and less secret as time goes by. Not only do these problems compromise the individual's privacy, but they also gradually destroy the usefulness of the information as an authenticator. When the information's value as an authenticator has been substantially degraded, another piece of (private) information will have to be chosen as a new basis for the private-secret authentication process. If this sequence of events is repeated enough times, many of the individual's private secrets will have been revealed to large numbers of other parties. The individual's privacy will have been put at considerable risk, as will the ability of other parties to authenticate the individual. Social Security numbers are an excellent case study in this phenomenon, as a recent incident involving the Princeton University admissions office illustrates. Both Princeton University and Yale University use Social Security numbers as student identifiers. Yale's admissions office used the Social Security number as an authentication key to allow students to access the status of their applications for admission via a Web browser. A member of the Princeton University admissions office staff discovered this and apparently used Social Security numbers obtained from the records of applicants to Princeton in order to access the Yale admissions Web site and learn about Yale's admissions decisions." _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Thu Mar 10 15:16:59 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA01010 for ; Thu, 10 Mar 2005 15:16:59 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9U8U-0004sn-Sx for speechsc-web-archive@ietf.org; Thu, 10 Mar 2005 15:19:59 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9U4l-0002d9-77; Thu, 10 Mar 2005 15:16:07 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9U4i-0002d4-2I for speechsc@megatron.ietf.org; Thu, 10 Mar 2005 15:16:05 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA00729 for ; Thu, 10 Mar 2005 15:16:02 -0500 (EST) Received: from salvelinus.brooktrout.com ([204.176.205.6]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9U7Z-0004pm-Mu for speechsc@ietf.org; Thu, 10 Mar 2005 15:19:02 -0500 Received: from nhmail2.needham.brooktrout.com (nhmail2.brooktrout.com [204.176.205.242]) by salvelinus.brooktrout.com (8.12.5/8.12.5) with ESMTP id j2AKC4dK019754; Thu, 10 Mar 2005 15:12:04 -0500 (EST) Received: by nhmail2.brooktrout.com with Internet Mail Service (5.5.2653.19) id ; Thu, 10 Mar 2005 15:08:06 -0500 Message-ID: From: Eric Burger To: "Shanmugham, Saravanan" , Skip Cave , Eric Burger , =?iso-8859-1?Q?Jes=FAs_Bernat_Vercher?= , Paolo.Baggia@LOQUENDO.COM, Dave Burke , jk@us.ibm.com, David R Oran Date: Thu, 10 Mar 2005 15:08:02 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) X-Spam-Score: 0.5 (/) X-Scan-Signature: 83e9494d829b08cc3f644ef6ac1b9bd4 Cc: "IETF SPEECHSC \(E-mail\)" Subject: [Speechsc] RE: Recording Functionality X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Content-Type: multipart/mixed; boundary="===============1947863092==" Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.6 (/) X-Scan-Signature: 025f8c5000216988bfe31585db759250 This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. --===============1947863092== Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C525AC.DC505242" This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01C525AC.DC505242 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Copying into the archives for posterity. -----Original Message----- From: Shanmugham, Saravanan [mailto:sarvi@cisco.com] Sent: Thursday, March 10, 2005 11:47 AM To: Skip Cave; Eric Burger; Jes=FAs Bernat Vercher; = Paolo.Baggia@LOQUENDO.COM; Dave Burke; jk@us.ibm.com; David R Oran Subject: RE: Recording Functionality Hi Skip, Looks like Jeff has addressed your "Message Control" question. As far as Recording functionality goes, it is to record the = incoming audio stream. It does not record a mix of both the incoming and = outgoing conversation. If a client wants that functionality, it would be the = responsibility of the client to involve a mixer somewhere to mix the 2 sides of the conversation and then feed it to the recording engine. =20 Sarvi _____ =20 From: Skip Cave [mailto:Skip.Cave@intervoice.com]=20 Sent: Wednesday, March 09, 2005 12:47 PM To: Eric Burger; Jes=FAs Bernat Vercher; Paolo.Baggia@LOQUENDO.COM; Shanmugham, Saravanan; Dave Burke; jk@us.ibm.com; David R Oran Subject: Recording Functionality All, =20 I have a question about the "Record" function in MRCP V2. Can it be = used to record both sides of a phone conversation, while TTS, ASR and users are = all interacting on other MRCP control channels?=20 =20 =20 Ellis K. 'Skip' Cave Chief Scientist, R&D Intervoice Inc. =20 ' 972.454.8800 (office) ' 214.460.4861 (cell) * skip.cave@intervoice.com =20 ------_=_NextPart_001_01C525AC.DC505242 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Copying into the archives for posterity.
-----Original Message-----
From: Shanmugham, = Saravanan=20 [mailto:sarvi@cisco.com]
Sent: Thursday, March 10, 2005 = 11:47=20 AM
To: Skip Cave; Eric Burger; Jes=FAs Bernat Vercher;=20 Paolo.Baggia@LOQUENDO.COM; Dave Burke; jk@us.ibm.com; David R=20 Oran
Subject: RE: Recording = Functionality

Hi Skip,
     Looks like Jeff = has=20 addressed your "Message Control" question.
     As far as = Recording=20 functionality goes, it is to record the incoming audio stream. It = does not=20 record a mix of both the incoming and outgoing=20 conversation.
     If a client wants = that=20 functionality, it would be the responsibility of the client to = involve a mixer=20 somewhere to mix the 2 sides of the conversation and then feed it to = the=20 recording engine.
 
Sarvi

From: Skip Cave=20 [mailto:Skip.Cave@intervoice.com]
Sent: Wednesday, March = 09, 2005=20 12:47 PM
To: Eric Burger; Jes=FAs Bernat Vercher;=20 Paolo.Baggia@LOQUENDO.COM; Shanmugham, Saravanan; Dave Burke; = jk@us.ibm.com;=20 David R Oran
Subject: Recording = Functionality

All,

 

I have a = question about=20 the "Record" function in MRCP V2. Can it be used to record both sides = of a=20 phone conversation, while TTS, ASR and users are all interacting on = other MRCP=20 control channels?

 

 

Ellis K. 'Skip' Cave

Chief Scientist,=20 R&D

Intervoice=20 Inc.

 

' = 972.454.8800=20 (office)

'=20 214.460.4861 (cell)

*=20 skip.cave@intervoice.com

 

------_=_NextPart_001_01C525AC.DC505242-- --===============1947863092== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline Content-Transfer-Encoding: 7bit _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc --===============1947863092==-- From speechsc-bounces@ietf.org Thu Mar 10 15:17:44 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA01264 for ; Thu, 10 Mar 2005 15:17:44 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9U9D-0004w8-SV for speechsc-web-archive@ietf.org; Thu, 10 Mar 2005 15:20:44 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9U4Q-0002ZJ-MO; Thu, 10 Mar 2005 15:15:46 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9U4N-0002Yf-Ap for speechsc@megatron.ietf.org; Thu, 10 Mar 2005 15:15:44 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA00660 for ; Thu, 10 Mar 2005 15:15:40 -0500 (EST) Received: from salvelinus.brooktrout.com ([204.176.205.6]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9U7E-0004ow-7i for speechsc@ietf.org; Thu, 10 Mar 2005 15:18:40 -0500 Received: from nhmail2.needham.brooktrout.com (nhmail2.brooktrout.com [204.176.205.242]) by salvelinus.brooktrout.com (8.12.5/8.12.5) with ESMTP id j2AKC4dL019755; Thu, 10 Mar 2005 15:12:05 -0500 (EST) Received: by nhmail2.brooktrout.com with Internet Mail Service (5.5.2653.19) id ; Thu, 10 Mar 2005 15:08:06 -0500 Message-ID: From: Eric Burger To: Skip Cave , Dave Burke , Eric Burger , =?iso-8859-1?Q?Jes=FAs_Bernat_Vercher?= , Paolo.Baggia@LOQUENDO.COM, Sarvi Shanmugham , jk@us.ibm.com, David R Oran Date: Thu, 10 Mar 2005 15:08:05 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) X-Spam-Score: 0.6 (/) X-Scan-Signature: 9a0f005dcc96c32c6d659bdf82d519da Cc: "IETF SPEECHSC \(E-mail\)" Subject: [Speechsc] RE: Message Control & Recording X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Content-Type: multipart/mixed; boundary="===============2090219291==" Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.6 (/) X-Scan-Signature: dd530fb5c702ad6eef32cae6dcb86145 This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. --===============2090219291== Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C525AC.DE429DBC" This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01C525AC.DE429DBC Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Copied to archive for posterity -----Original Message----- From: Skip Cave [mailto:Skip.Cave@intervoice.com] Sent: Thursday, March 10, 2005 12:59 PM To: Dave Burke; Eric Burger; Jes=FAs Bernat Vercher; Paolo.Baggia@LOQUENDO.COM; Sarvi Shanmugham; jk@us.ibm.com; David R = Oran Subject: Message Control & Recording Thanks to all of you, for the clarifications of the playback = volume/speed modification functionality. I was worried that the real-time audio = control functionality had been left out of the V2 spec, but it was in there, = just not obvious. Of course, the Voice Browser group still needs to expose = that real-time-control functionality in VXML. The current VXML FIA seems to prevent additional actions on messages that are already playing. =20 However, I'm still a bit unsure about the record function that I asked about. I don't necessarily want to combine both sides of a conversation = into a single monophonic file. Instead, I would like to record the received = audio from the caller, and the outbound audio from the Speech server/browser = into a single stereo file, with the system audio on one track, and the = caller audio on the other track. This scheme helps identify the source of = various anomalies in the tuning process. Of course, one would also like the = option to record the summed audio in a monophonic file, if needed.=20 =20 Assuming that I can route the two audio streams from different sources = into a single stereo RTP stream, will the record function allow the = recording of the complete call, or any portion of the call, while still performing = all the MRCP functions with the other MRCP resources concurrently? I am = really trying to understand if the recording function totally independent of = the ASR or TTS resources? Can I record audio at any time, regardless of the state of the ASR or TTS functions?=20 =20 --- Skip =20 =20 _____ =20 From: Dave Burke [mailto:david.burke@voxpilot.com]=20 Sent: Wednesday, March 09, 2005 6:40 PM To: Skip Cave; Eric Burger; Jes=FAs Bernat Vercher; = Paolo.Baggia@LOQUENDO.COM; Sarvi Shanmugham; jk@us.ibm.com; David R Oran Subject: Re: Message Control =20 Hi Skip, =20 The CONTROL verb along with the Prosody- header parameters (whose = semantics map to SSML attributes) MAY be used to affect an ongoing = SPEAK request. This allows you (assuming your speechsynth supports it) to = alter pitch, volume, rate etc. The Jump-Size header allows you to skip = around.=20 =20 -- Dave ----- Original Message -----=20 From: Skip Cave=20 To: Eric Burger ; Jes=FAs Bernat Vercher ; Paolo.Baggia@LOQUENDO.COM ; Sarvi Shanmugham ; Dave Burke = ; jk@us.ibm.com ; David R Oran = Sent: Wednesday, March 09, 2005 12:54 PM Subject: Message Control =20 Another question: =20 There is a general industry requirement to modify the playback of = messages during playback. The requirement is to change the volume, playback = speed, or skip around in the message, all under speech or DTMF commands from the = user. This is a very common requirement for voicemail applications, but is = used in many other types of applications as well.=20 =20 How does MRCP support this key requirement? I would expect that there = would be a set of MRCP V2 commands to invoke these functions during the play = of audio or TTS messages, but I don't see them in the spec.=20 =20 Skip =20 _____ =20 From: Skip Cave=20 Sent: Wednesday, March 09, 2005 2:47 PM To: 'Eric Burger'; 'Jes=FAs Bernat Vercher'; = 'Paolo.Baggia@LOQUENDO.COM' ; 'Sarvi Shanmugham'; 'Dave = Burke'; 'jk@us.ibm.com'; 'David R Oran' Subject: Recording Functionality =20 All, =20 I have a question about the "Record" function in MRCP V2. Can it be = used to record both sides of a phone conversation, while TTS, ASR and users are = all interacting on other MRCP control channels?=20 =20 =20 Ellis K. 'Skip' Cave Chief Scientist, R&D Intervoice Inc. =20 ' 972.454.8800 (office) ' 214.460.4861 (cell) * skip.cave@intervoice.com =20 ------_=_NextPart_001_01C525AC.DE429DBC Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Copied=20 to archive for posterity
-----Original Message-----
From: Skip Cave=20 [mailto:Skip.Cave@intervoice.com]
Sent: Thursday, March 10, = 2005=20 12:59 PM
To: Dave Burke; Eric Burger; Jes=FAs Bernat = Vercher;=20 Paolo.Baggia@LOQUENDO.COM; Sarvi Shanmugham; jk@us.ibm.com; David R=20 Oran
Subject: Message Control & = Recording

Thanks to = all of you,=20 for the clarifications of the playback volume/speed modification=20 functionality. I was worried that the real-time audio control = functionality=20 had been left out of the V2 spec, but it was in there, just not = obvious. =20 Of course, the Voice Browser group still needs to expose that=20 real-time-control functionality in VXML. The current VXML FIA seems = to prevent=20 additional actions on messages that are already=20 playing.

 

However, = I'm still a=20 bit unsure about the record function that I asked about. I don't = necessarily=20 want to combine both sides of a conversation into a single monophonic = file.=20 Instead, I would like to record the received audio from the caller, = and the=20 outbound audio from the Speech server/browser into a single stereo = file, with=20 the system audio on one track, and the caller audio on the other = track. This=20 scheme helps identify the source of various anomalies in the tuning = process.=20 Of course, one would also like the option to record the summed audio = in a=20 monophonic file, if needed.

 

Assuming = that I can=20 route the two audio streams from different sources into a single = stereo RTP=20 stream, will the record function allow the recording of the complete = call, or=20 any portion of the call, while still performing all the MRCP = functions with=20 the other MRCP resources concurrently? I am really trying to = understand if the=20 recording function totally independent of the ASR or TTS resources? = Can I=20 record audio at any time, regardless of the state of the ASR or TTS = functions?=20

 

---=20 Skip

 

 

From: Dave=20 Burke [mailto:david.burke@voxpilot.com]
Sent: Wednesday, March 09, = 2005 6:40=20 PM
To: Skip Cave; = Eric=20 Burger; Jes=FAs Bernat Vercher; Paolo.Baggia@LOQUENDO.COM; Sarvi = Shanmugham;=20 jk@us.ibm.com; David R Oran
Subject: Re: Message=20 Control

 

Hi=20 Skip,

 

The CONTROL verb along = with the=20 Prosody- header parameters (whose semantics map to SSML = <prosody>=20 attributes) MAY be used to affect an ongoing SPEAK request. This = allows you=20 (assuming your speechsynth supports it) to alter pitch, volume, rate = etc. The=20 Jump-Size header allows you to skip around.=20

 

--=20 Dave

----- Original = Message -----=20

From: Skip=20 Cave

To: Eric=20 Burger ; Jes=FAs Bernat=20 Vercher ; Paolo.Baggia@LOQUENDO.COM = ; Sarvi = Shanmugham ;=20 Dave Burke ; jk@us.ibm.com ; David R = Oran=20

Sent:=20 Wednesday, March 09, 2005 12:54 = PM

Subject: Message=20 Control

 

Another=20 question:

 

There is = a general=20 industry requirement to modify the playback of messages during = playback. The=20 requirement is to change the volume, playback speed, or skip around = in the=20 message, all under speech or DTMF commands from the user. This is a = very=20 common requirement for voicemail applications, but is used in many = other=20 types of applications as well.

 

How does = MRCP=20 support this key requirement? I would expect that there would be a = set of=20 MRCP V2 commands to invoke these functions during the play of audio = or TTS=20 messages, but I don't see them in the spec. =

 

Skip

 

From:=20 Skip=20 Cave =
Sent: Wednesday, March 09, = 2005 2:47=20 PM
To: 'Eric = Burger';=20 'Jes=FAs Bernat Vercher'; 'Paolo.Baggia@LOQUENDO.COM'<= /A>;=20 'Sarvi Shanmugham'; 'Dave Burke'; 'jk@us.ibm.com'; 'David R=20 Oran'
Subject: = Recording=20 Functionality

 

All,

 

I have a = question about=20 the "Record" function in MRCP V2. Can it be used to record both = sides of a=20 phone conversation, while TTS, ASR and users are all interacting on = other=20 MRCP control channels?

 

 

Ellis K. 'Skip' Cave

Chief Scientist,=20 R&D

Intervoice=20 Inc.

 

'=20 972.454.8800=20 (office)

'=20 214.460.4861 (cell)

*=20 skip.cave@intervoice.com

 

------_=_NextPart_001_01C525AC.DE429DBC-- --===============2090219291== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline Content-Transfer-Encoding: 7bit _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc --===============2090219291==-- From speechsc-bounces@ietf.org Thu Mar 10 23:31:14 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA14780 for ; Thu, 10 Mar 2005 23:31:13 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9bqt-0003UT-2D for speechsc-web-archive@ietf.org; Thu, 10 Mar 2005 23:34:19 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9bmq-0008BM-Pv; Thu, 10 Mar 2005 23:30:08 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1D9bmj-0008AF-VM for speechsc@megatron.ietf.org; Thu, 10 Mar 2005 23:30:02 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA14752 for ; Thu, 10 Mar 2005 23:29:59 -0500 (EST) Received: from salvelinus.brooktrout.com ([204.176.205.6]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1D9bpg-0003OU-2v for speechsc@ietf.org; Thu, 10 Mar 2005 23:33:04 -0500 Received: from nhmail2.needham.brooktrout.com (nhmail2.brooktrout.com [204.176.205.242]) by salvelinus.brooktrout.com (8.12.5/8.12.5) with ESMTP id j2B4OcfF024554 for ; Thu, 10 Mar 2005 23:24:38 -0500 (EST) Received: by nhmail2.brooktrout.com with Internet Mail Service (5.5.2653.19) id ; Thu, 10 Mar 2005 23:20:40 -0500 Message-ID: From: Eric Burger To: "IETF SPEECHSC (E-mail)" Date: Thu, 10 Mar 2005 23:20:37 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" X-Spam-Score: 0.0 (/) X-Scan-Signature: 093efd19b5f651b2707595638f6c4003 Subject: [Speechsc] Draft Meeting Minutes X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 2beba50d0fcdeee5f091c59f204d4365 Scribe: Keith Lantz Agenda: - Intro and agenda bashing - Push to last call for MRCPv2 (draft-ietf-speechsc-mrcpv2-06.txt) - Discussion of security issues on requirements (draft-ietf-speechsc-reqts-06.txt) - Wrap-up MRCPv2 (Sarvi Shanmugham) - At version 06; includes - Edits from last meeting - IANA considerations section - Security considerations section - Clarifications and comments discussed on alias through 2/21/05 - Baggia Paolo: Comments, Part 1 - Jeff Kusnitz: Grammar weights issue - Klaus Reifenrath: NLSML issues (partial) - Resolved issues NOT included in rev 6: - Baggia Paolo comments, part 2: - Editorial remarks - Negotiate a future alternative for NLSML, to allow the the client to optionally specify the alternative as a mime-type. - Cite the Semantic Interpretation spec, but do NOT add any additional text. - SDP changes: - Follow MMUSIC, per Magnus - Dropped SCTP for lack of SDP mechanism to address it; propose to address it independent of the Speechsc WG - IANA registrations: resource-types, method/event names, SDP proto values, SDP attributes, etc. - Dave and Sarvi to review with IANA tomorrow (3/9) - Open issues: - Do we want DEFINE-LEXICON support in the recognizer? Proposed by Baggia Paolo; Sarvi has no opinion. Will close offline via the list. - Security considerations (see below) - Should expect draft ready for last call within 2-3 weeks. SECURITY CONSIDERATIONS (Dave Oran) - Have been holding up requirements draft for over a year. - Principal concern (as voice, e.g., in CSTA Report): Using any biometric or any sort of speech recognition for speaker identification is a bad idea. - Privacy: main concern is potential for large scale theft of voiceprints and concomitant privacy loss - Security of protocols: usual questions about confidentiality, integrity and authentication/authorization - Threats/vulnerabilities of speech as a biometric - (Real) threats to SI/SV protocols: - External threats (i.e. by unauthorized parties) - Attacks can be foiled by well-understood means - Speechsc employs TLS encryption of the control channel SRTP encryption of the media channel Authentication/authorization of all elements ==> NOT an issue. - Internal attacks: stealing the voiceprint DB or compromising the the server system, including its keying material - Prudent to store the data encrypted to foil theft. - But Speechsc is a protocol standard, so not clear what we need to do anything about this ... But many people think we must. So ... - Do whatever is done with Radius! - Replay and impersonation attacks - Premise: Possession of a stolen voiceprint does NOT by itself enable impersonation, due to use of challenge-response protocols. - Stephan (?) disputed that premise, in light of the ability to construct suitable responses given enough voice data. He suggested adding knowledge base on top of challenge- response -- e.g. "What's the name of your favorite pet?" - Voiceprints of varying quality can be obtained by anything that can record your voice. So voiceprint used for SI/SV needs to be protected. - Question: Is it really useful to requires servers holding voiceprints to be more secure than those holding speech recordings, especially if those recordings have meta-data allowing the source to be identified (e.g. calling phone number, logged in user id)? - To be discussed on the list. - Other issues: - Implicit agreement that your identity can be ascertained by the participants? - What about a secure session with explicit anonymity? Is SI out of line with common expectations of privacy? - If indirect methods of id need to be thwarted, there are things like voice distorting devices that render the SI/SV impotent. - Stephan (?) expressed skepticism as to the need for Speechsc to address these issues. Unfortunately, per Dave, the IESG is equally skeptical about biometrics and has thus far required these issues be addressed. - Next steps: - Further discussion of the above issues on the list. - Work with Transport & Security ADs to resolve these issues. - If needed, establish an ongoing "security advisor" function to help get closure on both the requirements and the MRCPv2 spec. UPCOMING MILESTONES Apr 05 MRCPv2 WGLC Jun 05 Submit MRCPv2 to IESG _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Mon Mar 14 09:03:30 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA03192 for ; Mon, 14 Mar 2005 09:03:30 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DAqE0-00038q-MR for speechsc-web-archive@ietf.org; Mon, 14 Mar 2005 09:07:18 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DAq8A-00027E-CS; Mon, 14 Mar 2005 09:01:14 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DAq89-000278-81 for speechsc@megatron.ietf.org; Mon, 14 Mar 2005 09:01:13 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA02768 for ; Mon, 14 Mar 2005 09:01:11 -0500 (EST) Received: from sj-iport-2-in.cisco.com ([171.71.176.71] helo=sj-iport-2.cisco.com) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DAqBl-0002u0-EK for speechsc@ietf.org; Mon, 14 Mar 2005 09:04:59 -0500 Received: from sj-core-2.cisco.com (171.71.177.254) by sj-iport-2.cisco.com with ESMTP; 14 Mar 2005 06:16:11 -0800 Received: from imail.cisco.com (imail.cisco.com [128.107.200.91]) by sj-core-2.cisco.com (8.12.10/8.12.6) with ESMTP id j2EE0uq8023069; Mon, 14 Mar 2005 06:00:57 -0800 (PST) Received: from [10.32.245.156] (stealth-10-32-245-156.cisco.com [10.32.245.156]) by imail.cisco.com (8.12.11/8.12.10) with SMTP id j2EDsFe3010156; Mon, 14 Mar 2005 05:54:15 -0800 In-Reply-To: References: Mime-Version: 1.0 (Apple Message framework v619.2) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <495e5eb5661d70ce00c2efc255592024@cisco.com> Content-Transfer-Encoding: 7bit From: David R Oran Subject: Re: [Speechsc] Draft Meeting Minutes Date: Mon, 14 Mar 2005 09:00:55 -0500 To: Eric Burger X-Mailer: Apple Mail (2.619.2) IIM-SIG: v:"1.1"; h:"imail.cisco.com"; d:"cisco.com"; z:"home"; m:"krs"; t:"1110808456.76036"; x:"432200"; a:"rsa-sha1"; b:"nofws:4047"; e:"Iw=="; n:"sQYarK2E51MdcTiUqeif3F7cWdxIfoCiXhdfb9vD5ee/j0jXL15gbFxF2p" "XIweAblu0N6XAgK7k+wrbr7bQDJaCDqOmzqpRUBjIRQAXQ7NzadpmR3pUL6wxaRUtW+c43sl9jC" "50Qg1sXHpPjt8Y+Y16ioyQAQAdSunM4YhevURc="; s:"JnghNG4BvO0ztSTCQFJeG00CLndyVS0n2J6mZdx/F/N2XWIAWMSROfHN92T6L+gYdUJE2aDK" "gi8YXbk03u9qRUWlBLF2/ifeTClHMw4H0hfqgIxrUiL/EBetPwkDRwqmB7/9dMBM49nN/ME/RUa" "vFPIEtQxjzoKlidxQXGiFYBs="; c:"From: David R Oran "; c:"Subject: Re: [Speechsc] Draft Meeting Minutes"; c:"Date: Mon, 14 Mar 2005 09:00:55 -0500" IIM-VERIFY: s:"y"; v:"y"; r:"60"; h:"imail.cisco.com"; c:"message from imail.cisco.com verified; " X-Spam-Score: 0.0 (/) X-Scan-Signature: 29dc808194f5fb921c09d0040806d6eb Content-Transfer-Encoding: 7bit Cc: "IETF SPEECHSC \(E-mail\)" X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 3971661e40967acfc35f708dd5f33760 Content-Transfer-Encoding: 7bit Please send corrections/additions to me. I intend to submit the formal minutes at the end of this week. Dave, (co-chair). On Mar 10, 2005, at 11:20 PM, Eric Burger wrote: > Scribe: Keith Lantz > > > Agenda: > - Intro and agenda bashing > - Push to last call for MRCPv2 (draft-ietf-speechsc-mrcpv2-06.txt) > - Discussion of security issues on requirements > (draft-ietf-speechsc-reqts-06.txt) > - Wrap-up > > > MRCPv2 (Sarvi Shanmugham) > > - At version 06; includes > - Edits from last meeting > - IANA considerations section > - Security considerations section > - Clarifications and comments discussed on alias through 2/21/05 > - Baggia Paolo: Comments, Part 1 > - Jeff Kusnitz: Grammar weights issue > - Klaus Reifenrath: NLSML issues (partial) > > - Resolved issues NOT included in rev 6: > - Baggia Paolo comments, part 2: > - Editorial remarks > - Negotiate a future alternative for NLSML, to allow the > the client to optionally specify the alternative as > a mime-type. > - Cite the Semantic Interpretation spec, but do NOT add > any additional text. > - SDP changes: > - Follow MMUSIC, per Magnus > - Dropped SCTP for lack of SDP mechanism to address it; propose > to address it independent of the Speechsc WG > - IANA registrations: resource-types, method/event names, SDP > proto values, SDP attributes, etc. > - Dave and Sarvi to review with IANA tomorrow (3/9) > > - Open issues: > - Do we want DEFINE-LEXICON support in the recognizer? Proposed by > Baggia Paolo; Sarvi has no opinion. Will close offline via the > list. > - Security considerations (see below) > > - Should expect draft ready for last call within 2-3 weeks. > > > SECURITY CONSIDERATIONS (Dave Oran) > > - Have been holding up requirements draft for over a year. > > - Principal concern (as voice, e.g., in CSTA Report): Using any > biometric > or any sort of speech recognition for speaker identification is a bad > idea. > - Privacy: main concern is potential for large scale theft of > voiceprints and concomitant privacy loss > - Security of protocols: usual questions about confidentiality, > integrity and authentication/authorization > - Threats/vulnerabilities of speech as a biometric > > - (Real) threats to SI/SV protocols: > - External threats (i.e. by unauthorized parties) > - Attacks can be foiled by well-understood means > - Speechsc employs > TLS encryption of the control channel > SRTP encryption of the media channel > Authentication/authorization of all elements > ==> NOT an issue. > - Internal attacks: stealing the voiceprint DB or compromising the > the server system, including its keying material > - Prudent to store the data encrypted to foil theft. > - But Speechsc is a protocol standard, so not clear what > we need to do anything about this ... But many people > think we must. So ... > - Do whatever is done with Radius! > - Replay and impersonation attacks > - Premise: Possession of a stolen voiceprint does NOT by > itself enable impersonation, due to use of > challenge-response protocols. > - Stephan (?) disputed that premise, in light of the ability > to construct suitable responses given enough voice data. > He suggested adding knowledge base on top of challenge- > response -- e.g. "What's the name of your favorite pet?" > > - Voiceprints of varying quality can be obtained by anything that can > record your voice. So voiceprint used for SI/SV needs to be protected. > - Question: Is it really useful to requires servers holding > voiceprints to be more secure than those holding speech > recordings, > especially if those recordings have meta-data allowing the source > to be identified (e.g. calling phone number, logged in user id)? > - To be discussed on the list. > > - Other issues: > - Implicit agreement that your identity can be ascertained by the > participants? > - What about a secure session with explicit anonymity? Is SI > out of line with common expectations of privacy? > - If indirect methods of id need to be thwarted, there are things > like voice distorting devices that render the SI/SV impotent. > > - Stephan (?) expressed skepticism as to the need for Speechsc to > address > these issues. Unfortunately, per Dave, the IESG is equally skeptical > about > biometrics and has thus far required these issues be addressed. > > - Next steps: > - Further discussion of the above issues on the list. > - Work with Transport & Security ADs to resolve these issues. > - If needed, establish an ongoing "security advisor" function to > help > get closure on both the requirements and the MRCPv2 spec. > > > UPCOMING MILESTONES > > Apr 05 MRCPv2 WGLC > Jun 05 Submit MRCPv2 to IESG > > > > > > > > > > > _______________________________________________ > Speechsc mailing list > Speechsc@ietf.org > https://www1.ietf.org/mailman/listinfo/speechsc > David R. Oran Cisco Fellow Cisco Systems 7 Ladyslipper Lane Acton, MA 01720 USA Tel: +1 978 264 2048 Email: oran@cisco.com _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Tue Mar 15 15:34:49 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA16145 for ; Tue, 15 Mar 2005 15:34:49 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DBIoV-0005h8-Sl for speechsc-web-archive@ietf.org; Tue, 15 Mar 2005 15:38:53 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DBIiW-0007ab-3i; Tue, 15 Mar 2005 15:32:40 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DBIiU-0007aW-Km for speechsc@megatron.ietf.org; Tue, 15 Mar 2005 15:32:38 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA16015 for ; Tue, 15 Mar 2005 15:32:36 -0500 (EST) Received: from sj-iport-1-in.cisco.com ([171.71.176.70] helo=sj-iport-1.cisco.com) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DBImN-0005eA-6s for speechsc@ietf.org; Tue, 15 Mar 2005 15:36:40 -0500 Received: from sj-core-2.cisco.com (171.71.177.254) by sj-iport-1.cisco.com with ESMTP; 15 Mar 2005 12:27:48 -0800 X-BrightmailFiltered: true X-Brightmail-Tracker: AAAAAA== X-IronPort-AV: i="3.90,165,1107763200"; d="scan'208"; a="619657551:sNHT22104716" Received: from imail.cisco.com (imail.cisco.com [128.107.200.91]) by sj-core-2.cisco.com (8.12.10/8.12.6) with ESMTP id j2FKBB6d024297 for ; Tue, 15 Mar 2005 12:11:12 -0800 (PST) Received: from [10.32.245.156] (stealth-10-32-245-156.cisco.com [10.32.245.156]) by imail.cisco.com (8.12.11/8.12.10) with SMTP id j2FK4ScD023305 for ; Tue, 15 Mar 2005 12:04:29 -0800 Mime-Version: 1.0 (Apple Message framework v619.2) Content-Transfer-Encoding: 7bit Message-Id: <740d40f6a78063456b46d0e3bac66df3@cisco.com> Content-Type: text/plain; charset=US-ASCII; format=flowed To: IETF SPEECHSC (E-mail) From: David R Oran Date: Tue, 15 Mar 2005 15:11:12 -0500 X-Mailer: Apple Mail (2.619.2) IIM-SIG: v:"1.1"; h:"imail.cisco.com"; d:"cisco.com"; z:"home"; m:"krs"; t:"1110917069.306459"; x:"432200"; a:"rsa-sha1"; b:"nofws:1315"; e:"Iw=="; n:"sQYarK2E51MdcTiUqeif3F7cWdxIfoCiXhdfb9vD5ee/j0jXL15gbFxF2p" "XIweAblu0N6XAgK7k+wrbr7bQDJaCDqOmzqpRUBjIRQAXQ7NzadpmR3pUL6wxaRUtW+c43sl9jC" "50Qg1sXHpPjt8Y+Y16ioyQAQAdSunM4YhevURc="; s:"k2EzpVvxmH14viWQD3IZDjFkNQEjiiwOcf5t3ZlPDKxp3+JqXiWmPo7dVYSHxcemLEv74IkL" "tf2VfbZ895Z9TlRFai6r7iIEqm6oyXVSft9Pyd/otOLrWOitxgGaOzPnseh6kG/4lv5LGzwnNXm" "t9yqUcmcdssKadZt5Yc7HqOg="; c:"From: David R Oran "; c:"Subject: Status of security issue with speechsc requirements"; c:"Date: Tue, 15 Mar 2005 15:11:12 -0500" IIM-VERIFY: s:"y"; v:"y"; r:"60"; h:"imail.cisco.com"; c:"message from imail.cisco.com verified; " X-Spam-Score: 0.0 (/) X-Scan-Signature: bb8f917bb6b8da28fc948aeffb74aa17 Content-Transfer-Encoding: 7bit Subject: [Speechsc] Status of security issue with speechsc requirements X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 7baded97d9887f7a0c7e8a33c2e3ea1b Content-Transfer-Encoding: 7bit This is a short note just to update the working group on the status of this item, for which I provided an update at the working group meeting at IETF62 last week. Subsequent to the WG session, the chairs met with Jon Peterson (our area advisor) and Sam Hartman (Security area director) to discuss the issues. - We confirmed that the difficulty is strictly with the usage of speaker verification for authentication/authorization purposes and that the other requirements (particularly the usage scenarios for speaker identification) were not sticking points. - Towards the end of the discussion Sam counseled that Russ Housley (the other area director for security) was more vehement in his objections to the speaker verification functions than Sam was, and suggested it might be advisable to transfer responsibility for resolving this to him. Sam said he would be happy with any resolution that made Russ happy. Later in the week Jon met privately with Russ Housley. Russ said he would look at the document again and see if he could propose some text for the security consideration section that would satisfy his concerns and allow the document to move forward. I'll be following up with Jon in a week or two if we have not gotten an update. At the moment there are no action items for the author (me) or for the working group; we can put all of our energy into getting MRCPv2 ready for WG last call. Thanks to all of you who made the effort to attend the working group meeting in Minneapolis, and additional thanks to Dan Burnett who participated remotely. Dave Oran (speechsc co-chair) _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Tue Mar 22 08:27:56 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA11108 for ; Tue, 22 Mar 2005 08:27:56 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DDjVb-0002sU-RX for speechsc-web-archive@ietf.org; Tue, 22 Mar 2005 08:33:25 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DDjOl-0004lj-9Q; Tue, 22 Mar 2005 08:26:19 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DDjOk-0004lZ-1c for speechsc@megatron.ietf.org; Tue, 22 Mar 2005 08:26:18 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA10975 for ; Tue, 22 Mar 2005 08:26:16 -0500 (EST) Received: from mx1.scansoft.com ([198.71.64.81]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DDjU0-0002qj-2D for speechsc@ietf.org; Tue, 22 Mar 2005 08:31:45 -0500 Received: from pb-exchcon.pb.scansoft.com ([10.1.4.73]) by mx1.scansoft.com with InterScan Messaging Security Suite; Tue, 22 Mar 2005 08:37:51 -0500 Received: by pb-exchcon.pb.scansoft.com with Internet Mail Service (5.5.2653.19) id ; Tue, 22 Mar 2005 08:26:05 -0500 Message-ID: From: "Reifenrath, Klaus" To: "'Sarvi Shanmugham'" Date: Tue, 22 Mar 2005 08:25:56 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" X-Spam-Score: 0.0 (/) X-Scan-Signature: 7aafa0432175920a4b3e118e16c5cb64 Cc: speechsc@ietf.org Subject: [Speechsc] confidence range in NLSML X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 0a7aa2e6e558383d84476dc338324fab Hi Sarvi, the 06 draft is still not consistent regarding NLSML: DTD and Schema specify a range of 0.0 to 1.0 for interpretation confidences (and most examples are now using this range), while the text (page 79) and the examples on page 79 use a range of 0 to 100. In NLSML confidence always was an integer from 0 to 100. It may look sort of weird, because the MRCP header fields (e.g. Confidence-Threshold) that deal with confidences are floats from 0.0 to 1.0. On the one hand we decided not to change NLSML but at the same time choose floats for confidences in header fields with respect to EMMA that will soon replace NLSML. Regards, Klaus -----Original Message----- From: Sarvi Shanmugham [mailto:sarvi@cisco.com] Sent: Dienstag, 22. Februar 2005 05:24 To: Dan Burnett; David R. Oran; Eric Burger; speechsc@ietf.org Subject: [Speechsc] draft mrcpv2 -06 is out Hi, The -06 version of the draft was submitted last night. This addresses the following. 1. Baggia Paolo's review comments. 2. Klaus's feedback on NLSML addressed. 3. Jeff Kusnitz's concern for supporting grammar weights. - As described in the email discussions. 4. Some ABNF cleanup 5. The Secutiry considerations section 6. The IANA considerations section. - SCTP and SCTP+TLS SDP media type is not supported in this draft, and is defered to a separate specification. - All other necessary registrations have been addressed. Thanks, Sarvi _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Tue Mar 22 08:32:40 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA11681 for ; Tue, 22 Mar 2005 08:32:40 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DDjaB-00035e-Ad for speechsc-web-archive@ietf.org; Tue, 22 Mar 2005 08:38:08 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DDjSp-0005FH-Cd; Tue, 22 Mar 2005 08:30:31 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DDjSn-0005FC-Rv for speechsc@megatron.ietf.org; Tue, 22 Mar 2005 08:30:29 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA11471 for ; Tue, 22 Mar 2005 08:30:28 -0500 (EST) Received: from web14122.mail.yahoo.com ([66.163.171.113]) by ietf-mx.ietf.org with smtp (Exim 4.33) id 1DDjY4-0002wX-LM for speechsc@ietf.org; Tue, 22 Mar 2005 08:35:57 -0500 Received: (qmail 11346 invoked by uid 60001); 22 Mar 2005 13:30:27 -0000 Comment: DomainKeys? See http://antispam.yahoo.com/domainkeys DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; b=Cv45HncdtBFBeg5X5wzGZuAM5ljVNf0OE0BLcErtcvqkicycJ87x0BAFUyHd7ntlTZv3KBmfpmoDpFjhOX39S1+MOAEcPVo3vl8DkeI+1afodzCzS0b293S7wY01sbsdsFiKgEaS44Lam/IJBU326ExAPHcO9sOEstEg1s6OXcA= ; Message-ID: <20050322133027.11344.qmail@web14122.mail.yahoo.com> Received: from [63.193.248.104] by web14122.mail.yahoo.com via HTTP; Tue, 22 Mar 2005 05:30:27 PST Date: Tue, 22 Mar 2005 05:30:27 -0800 (PST) From: Dan Burnett To: "Reifenrath, Klaus" , "'Sarvi Shanmugham'" In-Reply-To: 6667 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Spam-Score: 3.0 (+++) X-Scan-Signature: d185fa790257f526fedfd5d01ed9c976 Cc: speechsc@ietf.org Subject: [Speechsc] Re: confidence range in NLSML X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 3.0 (+++) X-Scan-Signature: c3a18ef96977fc9bcc21a621cbf1174b We should be using 0.0 to 1.0 everywhere. -- dan --- "Reifenrath, Klaus" wrote: > Hi Sarvi, > > the 06 draft is still not consistent regarding > NLSML: > DTD and Schema specify a range of 0.0 to 1.0 for > interpretation > confidences (and most examples are now using this > range), > while the text (page 79) and the examples on page 79 > use a range of 0 to > 100. > > In NLSML confidence always was an integer from 0 to > 100. > > It may look sort of weird, because the MRCP header > fields (e.g. > Confidence-Threshold) > that deal with confidences are floats from 0.0 to > 1.0. On the one hand we > decided not > to change NLSML but at the same time choose floats > for confidences in header > fields > with respect to EMMA that will soon replace NLSML. > > Regards, > Klaus > > -----Original Message----- > From: Sarvi Shanmugham [mailto:sarvi@cisco.com] > Sent: Dienstag, 22. Februar 2005 05:24 > To: Dan Burnett; David R. Oran; Eric Burger; > speechsc@ietf.org > Subject: [Speechsc] draft mrcpv2 -06 is out > > > > > Hi, > The -06 version of the draft was submitted > last night. This > addresses the following. > > 1. Baggia Paolo's review comments. > 2. Klaus's feedback on NLSML addressed. > 3. Jeff Kusnitz's concern for supporting grammar > weights. > - As described in the email discussions. > 4. Some ABNF cleanup > 5. The Secutiry considerations section > 6. The IANA considerations section. > - SCTP and SCTP+TLS SDP media type is not > supported in this draft, > and is defered to a separate specification. > - All other necessary registrations have been > addressed. > > Thanks, > Sarvi > > > > _______________________________________________ > Speechsc mailing list > Speechsc@ietf.org > https://www1.ietf.org/mailman/listinfo/speechsc > __________________________________ Do you Yahoo!? Yahoo! Mail - now with 250MB free storage. Learn more. http://info.mail.yahoo.com/mail_250 _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Tue Mar 22 08:36:16 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA12046 for ; Tue, 22 Mar 2005 08:36:16 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DDjdg-0003Bi-Jm for speechsc-web-archive@ietf.org; Tue, 22 Mar 2005 08:41:44 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DDjXA-0005p8-3W; Tue, 22 Mar 2005 08:35:00 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DDjX8-0005p3-0j for speechsc@megatron.ietf.org; Tue, 22 Mar 2005 08:34:58 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA11884 for ; Tue, 22 Mar 2005 08:34:56 -0500 (EST) Received: from mx1.scansoft.com ([198.71.64.81]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DDjcP-000395-6g for speechsc@ietf.org; Tue, 22 Mar 2005 08:40:25 -0500 Received: from pb-exchcon.pb.scansoft.com ([10.1.4.73]) by mx1.scansoft.com with InterScan Messaging Security Suite; Tue, 22 Mar 2005 08:46:25 -0500 Received: by pb-exchcon.pb.scansoft.com with Internet Mail Service (5.5.2653.19) id ; Tue, 22 Mar 2005 08:34:39 -0500 Message-ID: From: "Reifenrath, Klaus" To: "'Dan Burnett'" , "'Sarvi Shanmugham'" Date: Tue, 22 Mar 2005 08:34:31 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable X-Spam-Score: 0.0 (/) X-Scan-Signature: 34d35111647d654d033d58d318c0d21a Content-Transfer-Encoding: quoted-printable Cc: speechsc@ietf.org Subject: [Speechsc] RE: confidence range in NLSML X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: e1b0e72ff1bbd457ceef31828f216a86 Content-Transfer-Encoding: quoted-printable But in case of NLSML that will break backwards compatibility, because = today NLSML uses 0 to 100! Klaus -----Original Message----- From: Dan Burnett [mailto:dan_burnett2000@yahoo.com] Sent: Dienstag, 22. M=E4rz 2005 14:30 To: Reifenrath, Klaus; 'Sarvi Shanmugham' Cc: speechsc@ietf.org; Dan Burnett Subject: Re: confidence range in NLSML We should be using 0.0 to 1.0 everywhere. -- dan --- "Reifenrath, Klaus" wrote: > Hi Sarvi, >=20 > the 06 draft is still not consistent regarding > NLSML: > DTD and Schema specify a range of 0.0 to 1.0 for > interpretation > confidences (and most examples are now using this > range),=20 > while the text (page 79) and the examples on page 79 > use a range of 0 to > 100. >=20 > In NLSML confidence always was an integer from 0 to > 100. >=20 > It may look sort of weird, because the MRCP header > fields (e.g. > Confidence-Threshold) > that deal with confidences are floats from 0.0 to > 1.0. On the one hand we > decided not=20 > to change NLSML but at the same time choose floats > for confidences in header > fields=20 > with respect to EMMA that will soon replace NLSML. >=20 > Regards, > Klaus >=20 > -----Original Message----- > From: Sarvi Shanmugham [mailto:sarvi@cisco.com] > Sent: Dienstag, 22. Februar 2005 05:24 > To: Dan Burnett; David R. Oran; Eric Burger; > speechsc@ietf.org > Subject: [Speechsc] draft mrcpv2 -06 is out >=20 >=20 >=20 >=20 > Hi, > The -06 version of the draft was submitted > last night. This=20 > addresses the following. >=20 > 1. Baggia Paolo's review comments. > 2. Klaus's feedback on NLSML addressed. > 3. Jeff Kusnitz's concern for supporting grammar > weights. > - As described in the email discussions. > 4. Some ABNF cleanup > 5. The Secutiry considerations section > 6. The IANA considerations section. > - SCTP and SCTP+TLS SDP media type is not > supported in this draft,=20 > and is defered to a separate specification. > - All other necessary registrations have been > addressed. >=20 > Thanks, > Sarvi >=20 >=20 >=20 > _______________________________________________ > Speechsc mailing list > Speechsc@ietf.org > https://www1.ietf.org/mailman/listinfo/speechsc >=20 =09 __________________________________=20 Do you Yahoo!?=20 Yahoo! Mail - now with 250MB free storage. Learn more.=20 http://info.mail.yahoo.com/mail_250 _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Tue Mar 22 11:14:51 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA29527 for ; Tue, 22 Mar 2005 11:14:51 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DDm7B-0000mF-PO for speechsc-web-archive@ietf.org; Tue, 22 Mar 2005 11:20:22 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DDm1C-000682-Kl; Tue, 22 Mar 2005 11:14:10 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DDm1B-000676-49 for speechsc@megatron.ietf.org; Tue, 22 Mar 2005 11:14:09 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA29440 for ; Tue, 22 Mar 2005 11:14:07 -0500 (EST) Received: from sj-iport-3-in.cisco.com ([171.71.176.72] helo=sj-iport-3.cisco.com) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DDm6S-0000kl-Gr for speechsc@ietf.org; Tue, 22 Mar 2005 11:19:37 -0500 Received: from sj-core-1.cisco.com (171.71.177.237) by sj-iport-3.cisco.com with ESMTP; 22 Mar 2005 08:13:13 -0800 X-IronPort-AV: i="3.91,111,1110182400"; d="scan'208"; a="238835364:sNHT3682230382" Received: from vtg-um-e2k1.sj21ad.cisco.com (vtg-um-e2k1.cisco.com [171.70.93.55]) by sj-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id j2MGD8Dr005412; Tue, 22 Mar 2005 08:13:08 -0800 (PST) content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable X-MimeOLE: Produced By Microsoft Exchange V6.0.6603.0 Date: Tue, 22 Mar 2005 08:13:13 -0800 Message-ID: <6677B3346233B94EBB11C060935101200A4DE5EB@vtg-um-e2k1.sj21ad.cisco.com> Thread-Topic: confidence range in NLSML Thread-Index: AcUu4/XijYo4x6S4Qi60idv9tmaBnAAFXjYA From: "Shanmugham, Saravanan" To: "Reifenrath, Klaus" , "Dan Burnett" X-Spam-Score: 0.0 (/) X-Scan-Signature: 6e922792024732fb1bb6f346e63517e4 Content-Transfer-Encoding: quoted-printable Cc: speechsc@ietf.org Subject: [Speechsc] RE: confidence range in NLSML X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 9a2be21919e71dc6faef12b370c4ecf5 Content-Transfer-Encoding: quoted-printable I think I agree with Dan.=20 If we think 0.0 - 1.0 is the right way long term, I think that's what we = should do. This is a new protocol, though we have leveraged a lot of work from the = MRCPv1. So there should not be any backward compatibility issues. Also the NLSML markup compatibility should also not be a problem, = considering that we have officially only defined NLSML in this document = and this would be the official version 1.0 of NLSML.=20 My 2c, Sarvi -----Original Message----- From: Reifenrath, Klaus [mailto:Klaus.Reifenrath@Scansoft.com]=20 Sent: Tuesday, March 22, 2005 5:35 AM To: 'Dan Burnett'; Shanmugham, Saravanan Cc: speechsc@ietf.org Subject: RE: confidence range in NLSML But in case of NLSML that will break backwards compatibility, because = today NLSML uses 0 to 100! Klaus -----Original Message----- From: Dan Burnett [mailto:dan_burnett2000@yahoo.com] Sent: Dienstag, 22. M=E4rz 2005 14:30 To: Reifenrath, Klaus; 'Sarvi Shanmugham' Cc: speechsc@ietf.org; Dan Burnett Subject: Re: confidence range in NLSML We should be using 0.0 to 1.0 everywhere. -- dan --- "Reifenrath, Klaus" wrote: > Hi Sarvi, >=20 > the 06 draft is still not consistent regarding > NLSML: > DTD and Schema specify a range of 0.0 to 1.0 for interpretation=20 > confidences (and most examples are now using this range), while the=20 > text (page 79) and the examples on page 79 use a range of 0 to 100. >=20 > In NLSML confidence always was an integer from 0 to 100. >=20 > It may look sort of weird, because the MRCP header fields (e.g. > Confidence-Threshold) > that deal with confidences are floats from 0.0 to 1.0. On the one hand = > we decided not to change NLSML but at the same time choose floats for=20 > confidences in header fields with respect to EMMA that will soon=20 > replace NLSML. >=20 > Regards, > Klaus >=20 > -----Original Message----- > From: Sarvi Shanmugham [mailto:sarvi@cisco.com] > Sent: Dienstag, 22. Februar 2005 05:24 > To: Dan Burnett; David R. Oran; Eric Burger; speechsc@ietf.org > Subject: [Speechsc] draft mrcpv2 -06 is out >=20 >=20 >=20 >=20 > Hi, > The -06 version of the draft was submitted last night. This=20 > addresses the following. >=20 > 1. Baggia Paolo's review comments. > 2. Klaus's feedback on NLSML addressed. > 3. Jeff Kusnitz's concern for supporting grammar weights. > - As described in the email discussions. > 4. Some ABNF cleanup > 5. The Secutiry considerations section 6. The IANA considerations=20 > section. > - SCTP and SCTP+TLS SDP media type is not supported in this=20 > draft, and is defered to a separate specification. > - All other necessary registrations have been addressed. >=20 > Thanks, > Sarvi >=20 >=20 >=20 > _______________________________________________ > Speechsc mailing list > Speechsc@ietf.org > https://www1.ietf.org/mailman/listinfo/speechsc >=20 =09 __________________________________ Do you Yahoo!?=20 Yahoo! Mail - now with 250MB free storage. Learn more.=20 http://info.mail.yahoo.com/mail_250 _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Tue Mar 22 11:43:06 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA03420 for ; Tue, 22 Mar 2005 11:43:06 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DDmYV-0002Ci-Hk for speechsc-web-archive@ietf.org; Tue, 22 Mar 2005 11:48:36 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DDmON-0001DO-15; Tue, 22 Mar 2005 11:38:07 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DDmOL-0001DJ-S4 for speechsc@megatron.ietf.org; Tue, 22 Mar 2005 11:38:05 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA02614 for ; Tue, 22 Mar 2005 11:38:03 -0500 (EST) Received: from mx1.scansoft.com ([198.71.64.81]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DDmTd-0001qt-FJ for speechsc@ietf.org; Tue, 22 Mar 2005 11:43:34 -0500 Received: from pb-exchcon.pb.scansoft.com ([10.1.4.73]) by mx1.scansoft.com with InterScan Messaging Security Suite; Tue, 22 Mar 2005 11:49:37 -0500 Received: by pb-exchcon.pb.scansoft.com with Internet Mail Service (5.5.2653.19) id ; Tue, 22 Mar 2005 11:37:51 -0500 Message-ID: From: "Reifenrath, Klaus" To: "'Shanmugham, Saravanan'" , Dan Burnett Date: Tue, 22 Mar 2005 11:36:28 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable X-Spam-Score: 0.0 (/) X-Scan-Signature: 5d7a7e767f20255fce80fa0b77fb2433 Content-Transfer-Encoding: quoted-printable Cc: speechsc@ietf.org Subject: [Speechsc] RE: confidence range in NLSML X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: b22590c27682ace61775ee7b453b40d3 Content-Transfer-Encoding: quoted-printable We decided at the San Diego F2F not to change NLSML, which by that time = was already a de facto standard, despite the fact that we knew of = several(!) shortcomings! I think we should stick to this decision to avoid lengthy discussion on all the other issues. Klaus -----Original Message----- From: Shanmugham, Saravanan [mailto:sarvi@cisco.com] Sent: Dienstag, 22. M=E4rz 2005 17:13 To: Reifenrath, Klaus; Dan Burnett Cc: speechsc@ietf.org Subject: RE: confidence range in NLSML I think I agree with Dan.=20 If we think 0.0 - 1.0 is the right way long term, I think that's what = we should do. This is a new protocol, though we have leveraged a lot of work from the MRCPv1. So there should not be any backward compatibility issues. Also the NLSML markup compatibility should also not be a problem, considering that we have officially only defined NLSML in this document = and this would be the official version 1.0 of NLSML.=20 My 2c, Sarvi -----Original Message----- From: Reifenrath, Klaus [mailto:Klaus.Reifenrath@Scansoft.com]=20 Sent: Tuesday, March 22, 2005 5:35 AM To: 'Dan Burnett'; Shanmugham, Saravanan Cc: speechsc@ietf.org Subject: RE: confidence range in NLSML But in case of NLSML that will break backwards compatibility, because = today NLSML uses 0 to 100! Klaus -----Original Message----- From: Dan Burnett [mailto:dan_burnett2000@yahoo.com] Sent: Dienstag, 22. M=E4rz 2005 14:30 To: Reifenrath, Klaus; 'Sarvi Shanmugham' Cc: speechsc@ietf.org; Dan Burnett Subject: Re: confidence range in NLSML We should be using 0.0 to 1.0 everywhere. -- dan --- "Reifenrath, Klaus" wrote: > Hi Sarvi, >=20 > the 06 draft is still not consistent regarding > NLSML: > DTD and Schema specify a range of 0.0 to 1.0 for interpretation=20 > confidences (and most examples are now using this range), while the=20 > text (page 79) and the examples on page 79 use a range of 0 to 100. >=20 > In NLSML confidence always was an integer from 0 to 100. >=20 > It may look sort of weird, because the MRCP header fields (e.g. > Confidence-Threshold) > that deal with confidences are floats from 0.0 to 1.0. On the one = hand=20 > we decided not to change NLSML but at the same time choose floats for = > confidences in header fields with respect to EMMA that will soon=20 > replace NLSML. >=20 > Regards, > Klaus >=20 > -----Original Message----- > From: Sarvi Shanmugham [mailto:sarvi@cisco.com] > Sent: Dienstag, 22. Februar 2005 05:24 > To: Dan Burnett; David R. Oran; Eric Burger; speechsc@ietf.org > Subject: [Speechsc] draft mrcpv2 -06 is out >=20 >=20 >=20 >=20 > Hi, > The -06 version of the draft was submitted last night. This=20 > addresses the following. >=20 > 1. Baggia Paolo's review comments. > 2. Klaus's feedback on NLSML addressed. > 3. Jeff Kusnitz's concern for supporting grammar weights. > - As described in the email discussions. > 4. Some ABNF cleanup > 5. The Secutiry considerations section 6. The IANA considerations=20 > section. > - SCTP and SCTP+TLS SDP media type is not supported in this=20 > draft, and is defered to a separate specification. > - All other necessary registrations have been addressed. >=20 > Thanks, > Sarvi >=20 >=20 >=20 > _______________________________________________ > Speechsc mailing list > Speechsc@ietf.org > https://www1.ietf.org/mailman/listinfo/speechsc >=20 =09 __________________________________ Do you Yahoo!?=20 Yahoo! Mail - now with 250MB free storage. Learn more.=20 http://info.mail.yahoo.com/mail_250 _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc From speechsc-bounces@ietf.org Wed Mar 23 22:21:11 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA26375 for ; Wed, 23 Mar 2005 22:21:10 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DEIzs-0001dB-A6 for speechsc-web-archive@ietf.org; Wed, 23 Mar 2005 22:27:00 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DEIrQ-0001Jt-6Y; Wed, 23 Mar 2005 22:18:16 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DEIrO-0001Ji-LL for speechsc@megatron.ietf.org; Wed, 23 Mar 2005 22:18:14 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA26164 for ; Wed, 23 Mar 2005 22:18:12 -0500 (EST) Received: from sj-iport-3-in.cisco.com ([171.71.176.72] helo=sj-iport-3.cisco.com) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DEIwz-0001Xc-HX for speechsc@ietf.org; Wed, 23 Mar 2005 22:24:02 -0500 Received: from sj-core-2.cisco.com (171.71.177.254) by sj-iport-3.cisco.com with ESMTP; 23 Mar 2005 19:18:05 -0800 X-IronPort-AV: i="3.91,116,1110182400"; d="scan'208,217"; a="239829313:sNHT48258364" Received: from mira-sjc5-c.cisco.com (IDENT:mirapoint@mira-sjc5-c.cisco.com [171.71.163.17]) by sj-core-2.cisco.com (8.12.10/8.12.6) with ESMTP id j2O3HwgE014141; Wed, 23 Mar 2005 19:17:58 -0800 (PST) Received: from mgindi-wxp.cisco.com ([128.107.151.77]) by mira-sjc5-c.cisco.com (MOS 3.4.5-GR) with ESMTP id BFB09214; Wed, 23 Mar 2005 19:18:01 -0800 (PST) Message-Id: <4.3.2.7.2.20050323191703.01c58e68@mira-sjc5-c.cisco.com> X-Sender: mgindi@mira-sjc5-c.cisco.com X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Wed, 23 Mar 2005 19:18:30 -0800 To: From: Madhu Gindi Mime-Version: 1.0 X-Spam-Score: 0.0 (/) X-Scan-Signature: 6cca30437e2d04f45110f2ff8dc1b1d5 Cc: jyao@cisco.com Subject: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCP V2 X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Content-Type: multipart/mixed; boundary="===============0922929769==" Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.0 (/) X-Scan-Signature: 8de5f93cb2b4e3bee75302e9eacc33db --===============0922929769== Content-Type: multipart/alternative; boundary="=====================_1155331828==_.ALT" --=====================_1155331828==_.ALT Content-Type: text/plain; charset="us-ascii" I am forwarding on Jenny's behalf since Jenny's message bounced back with error. /madhu ==================================== >Reply-To: >From: "Jenny Yao \(jyao\)" >To: >Cc: "'Sunitha Krishna'" , , > "'Madhu Gindi'" >Subject: Question for "accept" in VoiceXML 2.0 using MRCP V2 >Date: Wed, 23 Mar 2005 16:36:04 -0800 > >Our VoiceXML 2.0 interpreter uses MRCP to communicate with ASR/TTS server. In VoiceXML 2.0, there is an attribute "accept" for "menu" and "choice" elements. "accept" can be set to "exact" or "approximate" . If the accept attribute is "exact" then the user must say the entire phrase in the same order in which they occur in the choice phrase.If the accept attribute is "approximate", then the choice may be matched when a user says a subphrase of the expression . We don't see MRCP V2 draft address this capability. We would like to know if there is any solution other than adding this to MRCP V2 draft. If not, we would like to propose adding this feature in MRCP V2. > >Thanks. > >Jenny > >--- >Jenny Yao >Cisco Systems >Mail Stop SJ-22/2, 170 W. Tasman Drive, San Jose, CA 95134 >Voice: (408)525-3879 ________________________________________________________________ Madhu Gindi Software Development Manager MCEBU ph: 408-525-5571 --=====================_1155331828==_.ALT Content-Type: text/html; charset="us-ascii" I am forwarding on Jenny's behalf since Jenny's message bounced back with error.

/madhu
====================================
Reply-To: <jyao@cisco.com>
From: "Jenny Yao \(jyao\)" <jyao@cisco.com>
To: <speechsc@ietf.org>
Cc: "'Sunitha Krishna'" <sunkrish@cisco.com>, <sarvi@cisco.com>,
        "'Madhu Gindi'" <mgindi@cisco.com>
Subject: Question for "accept" in VoiceXML 2.0 using MRCP V2
Date: Wed, 23 Mar 2005 16:36:04 -0800

Our VoiceXML 2.0 interpreter uses MRCP to communicate with ASR/TTS server. In VoiceXML 2.0, there is an attribute "accept" for "menu" and "choice" elements. "accept" can be set to "exact" or "approximate" . If the accept attribute is "exact" then the user must say the entire phrase in the same order in which they occur in the choice phrase.If the accept attribute is "approximate", then the choice may be matched when a user says a subphrase of the expression . We don't see MRCP V2 draft address this capability. We would like to know if there is any solution other than adding this to MRCP V2 draft. If not, we would like to propose adding this feature in MRCP V2.
 
Thanks.
 
Jenny
 
---
Jenny Yao
Cisco Systems
Mail Stop SJ-22/2, 170 W. Tasman Drive, San Jose, CA 95134
Voice: (408)525-3879


________________________________________________________________
Madhu Gindi  Software Development Manager MCEBU ph: 408-525-5571 --=====================_1155331828==_.ALT-- --===============0922929769== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline Content-Transfer-Encoding: 7bit _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc --===============0922929769==-- From speechsc-bounces@ietf.org Thu Mar 24 08:46:27 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA12438 for ; Thu, 24 Mar 2005 08:46:26 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DESl2-0003fk-Lv for speechsc-web-archive@ietf.org; Thu, 24 Mar 2005 08:52:21 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DESdc-0000o9-Oe; Thu, 24 Mar 2005 08:44:40 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DESdY-0000nb-Ea for speechsc@megatron.ietf.org; Thu, 24 Mar 2005 08:44:38 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA12222 for ; Thu, 24 Mar 2005 08:44:34 -0500 (EST) Received: from fw01.db01.voxpilot.com ([212.17.54.82] helo=mail.voxpilot.com) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DESjC-0003aJ-Ra for speechsc@ietf.org; Thu, 24 Mar 2005 08:50:28 -0500 Received: from daburkewxp (unknown [195.25.185.61]) by mail.voxpilot.com (Postfix) with ESMTP id BD36221403C; Thu, 24 Mar 2005 13:44:28 +0000 (GMT) Message-ID: <02c601c53077$98a15ac0$0601a8c0@db01.voxpilot.com> From: "Dave Burke" To: , "Madhu Gindi" References: <4.3.2.7.2.20050323191703.01c58e68@mira-sjc5-c.cisco.com> Subject: Re: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCP V2 Date: Thu, 24 Mar 2005 14:44:25 +0100 MIME-Version: 1.0 X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2900.2180 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180 X-Spam-Score: 0.5 (/) X-Scan-Signature: ccfb4541e989aa743998098cd315d0fd Cc: jyao@cisco.com X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Content-Type: multipart/mixed; boundary="===============1885060098==" Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.5 (/) X-Scan-Signature: a0534e6179a1e260079328e8b03c7901 This is a multi-part message in MIME format. --===============1885060098== Content-Type: multipart/alternative; boundary="----=_NextPart_000_02BF_01C5307F.F9946900" This is a multi-part message in MIME format. ------=_NextPart_000_02BF_01C5307F.F9946900 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable The "accept" attribute mechanism described in VoiceXML calls for the = generation of a grammar which will be language and platform dependent.=20 The content of SRGS grammars is orthogonal to the MRCP protocol and does = not belong in it. -- Dave ----- Original Message -----=20 From: Madhu Gindi=20 To: speechsc@ietf.org=20 Cc: jyao@cisco.com=20 Sent: Thursday, March 24, 2005 4:18 AM Subject: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using = MRCP V2 I am forwarding on Jenny's behalf since Jenny's message bounced back = with error.=20 /madhu = =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D Reply-To: From: "Jenny Yao \(jyao\)" To: Cc: "'Sunitha Krishna'" , , "'Madhu Gindi'" Subject: Question for "accept" in VoiceXML 2.0 using MRCP V2 Date: Wed, 23 Mar 2005 16:36:04 -0800 Our VoiceXML 2.0 interpreter uses MRCP to communicate with ASR/TTS = server. In VoiceXML 2.0, there is an attribute "accept" for "menu" and = "choice" elements. "accept" can be set to "exact" or "approximate" . If = the accept attribute is "exact" then the user must say the entire phrase = in the same order in which they occur in the choice phrase.If the accept = attribute is "approximate", then the choice may be matched when a user = says a subphrase of the expression . We don't see MRCP V2 draft address = this capability. We would like to know if there is any solution other = than adding this to MRCP V2 draft. If not, we would like to propose = adding this feature in MRCP V2.=20 =20 Thanks. =20 Jenny=20 =20 --- Jenny Yao Cisco Systems Mail Stop SJ-22/2, 170 W. Tasman Drive, San Jose, CA 95134 Voice: (408)525-3879 ________________________________________________________________ Madhu Gindi Software Development Manager MCEBU ph: 408-525-5571=20 -------------------------------------------------------------------------= ----- _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc ------=_NextPart_000_02BF_01C5307F.F9946900 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
The "accept" attribute mechanism = described in=20 VoiceXML calls for the generation of a grammar which will = be language=20 and platform dependent.
 
The content of SRGS grammars is = orthogonal to the=20 MRCP protocol and does not belong in it.
 
-- Dave
----- Original Message -----
From:=20 Madhu = Gindi=20
Sent: Thursday, March 24, 2005 = 4:18=20 AM
Subject: [Speechsc] Fwd: = Question for=20 "accept" in VoiceXML 2.0 using MRCP V2

I am forwarding on Jenny's behalf since Jenny's message = bounced=20 back with error. =

/madhu
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
Reply-To: <jyao@cisco.com>
From: = "Jenny Yao=20 \(jyao\)" <jyao@cisco.com>
To:=20 <speechsc@ietf.org>
Cc:=20 "'Sunitha Krishna'" <sunkrish@cisco.com>, = <sarvi@cisco.com>,
  &= nbsp;    =20 "'Madhu Gindi'" <mgindi@cisco.com>
Subject: = Question=20 for "accept" in VoiceXML 2.0 using MRCP V2
Date: Wed, 23 Mar 2005 = 16:36:04 -0800

Our VoiceXML 2.0 = interpreter=20 uses MRCP to communicate with ASR/TTS server. In VoiceXML 2.0, there = is an=20 attribute "accept" for "menu" and "choice" elements. "accept" can be = set to=20 "exact" or "approximate" . If the accept attribute is "exact" then = the user=20 must say the entire phrase in the same order in which they occur in = the=20 choice phrase.If the accept attribute is "approximate", then the = choice may=20 be matched when a user says a subphrase of the expression . We don't = see=20 MRCP V2 draft address this capability. We would like to know if = there is any=20 solution other than adding this to MRCP V2 draft. If not, we would = like to=20 propose adding this feature in MRCP V2.
 
Thanks.
 
Jenny=20
 
---
Jenny Yao
Cisco Systems
Mail Stop SJ-22/2, 170 W. = Tasman Drive,=20 San Jose, CA 95134
Voice: = (408)525-3879


=
________________________________________________________________Madhu=20 Gindi  Software Development Manager MCEBU ph: 408-525-5571=20

_______________________________________________
Speechsc = mailing=20 = list
Speechsc@ietf.org
https://www1.ietf.org/mailman/listinfo/speec= hsc
------=_NextPart_000_02BF_01C5307F.F9946900-- --===============1885060098== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline Content-Transfer-Encoding: 7bit _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc --===============1885060098==-- From speechsc-bounces@ietf.org Thu Mar 24 11:42:20 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA01621 for ; Thu, 24 Mar 2005 11:42:20 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DEVVI-0001Ae-Kr for speechsc-web-archive@ietf.org; Thu, 24 Mar 2005 11:48:17 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DEVOc-0000TL-E4; Thu, 24 Mar 2005 11:41:22 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DEVOb-0000T3-30 for speechsc@megatron.ietf.org; Thu, 24 Mar 2005 11:41:21 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA01539 for ; Thu, 24 Mar 2005 11:41:18 -0500 (EST) Received: from sj-iport-5.cisco.com ([171.68.10.87]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DEVUI-00019L-SW for speechsc@ietf.org; Thu, 24 Mar 2005 11:47:15 -0500 Received: from sj-core-3.cisco.com (171.68.223.137) by sj-iport-5.cisco.com with ESMTP; 24 Mar 2005 08:41:11 -0800 Received: from vtg-um-e2k1.sj21ad.cisco.com (vtg-um-e2k1.cisco.com [171.70.93.55]) by sj-core-3.cisco.com (8.12.10/8.12.6) with ESMTP id j2OGf6gU029651; Thu, 24 Mar 2005 08:41:09 -0800 (PST) content-class: urn:content-classes:message MIME-Version: 1.0 X-MimeOLE: Produced By Microsoft Exchange V6.0.6603.0 Subject: RE: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCPV2 Date: Thu, 24 Mar 2005 08:41:13 -0800 Message-ID: <6677B3346233B94EBB11C060935101200A58D949@vtg-um-e2k1.sj21ad.cisco.com> Thread-Topic: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCPV2 Thread-Index: AcUwd+MuPnkm6N/7TSiF7nX2YdvwPwAFy3mA From: "Shanmugham, Saravanan" To: "Dave Burke" , , "Madhu Gindi" X-Spam-Score: 0.9 (/) X-Scan-Signature: 83867a50fd8f547996ccdaf89af24437 Cc: jyao@cisco.com X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Content-Type: multipart/mixed; boundary="===============1115288866==" Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.9 (/) X-Scan-Signature: 9cc83ac38bbbabacbf00f656311dd8d8 This is a multi-part message in MIME format. --===============1115288866== content-class: urn:content-classes:message Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C53090.4AB2C4DC" This is a multi-part message in MIME format. ------_=_NextPart_001_01C53090.4AB2C4DC Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable This is assuming the MRCP client will be generating the different grammar options for approximate. The grammar that needs to be generated for the approximate case is language dependent. As the system is much of the language intellegience and capbilities is on the server side.=20 =20 So would the system benefit from an accept=3Dexact/approximate header, that allows the recogniton/grammar engine on the server to use its language intelligence to render an equivalent approximate grammar? =20 In implementations such as a VoiceXML browser that depends on back end MRCP server, there may not be much language intelligence on the server to do this generation for different languages. =20 And if we do decide to provide this header, the question is where? Regular header in the DEFINE-GRAMMAR or RECOGNIZE? Or a Content header like Content-Match similar to the Content-Weight that is being proposed. I am thinking the later. =20 any thoughts, Sarvi =20 _____ =20 From: speechsc-bounces@ietf.org [mailto:speechsc-bounces@ietf.org] On Behalf Of Dave Burke Sent: Thursday, March 24, 2005 5:44 AM To: speechsc@ietf.org; Madhu Gindi Cc: jyao@cisco.com Subject: Re: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCPV2 The "accept" attribute mechanism described in VoiceXML calls for the generation of a grammar which will be language and platform dependent.=20 =20 The content of SRGS grammars is orthogonal to the MRCP protocol and does not belong in it. =20 -- Dave ----- Original Message -----=20 From: Madhu Gindi =20 To: speechsc@ietf.org=20 Cc: jyao@cisco.com=20 Sent: Thursday, March 24, 2005 4:18 AM Subject: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCP V2 I am forwarding on Jenny's behalf since Jenny's message bounced back with error.=20 =09 /madhu = =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D =09 Reply-To: From: "Jenny Yao \(jyao\)" To: Cc: "'Sunitha Krishna'" , , "'Madhu Gindi'" Subject: Question for "accept" in VoiceXML 2.0 using MRCP V2 Date: Wed, 23 Mar 2005 16:36:04 -0800 =09 Our VoiceXML 2.0 interpreter uses MRCP to communicate with ASR/TTS server. In VoiceXML 2.0, there is an attribute "accept" for "menu" and "choice" elements. "accept" can be set to "exact" or "approximate" . If the accept attribute is "exact" then the user must say the entire phrase in the same order in which they occur in the choice phrase.If the accept attribute is "approximate", then the choice may be matched when a user says a subphrase of the expression . We don't see MRCP V2 draft address this capability. We would like to know if there is any solution other than adding this to MRCP V2 draft. If not, we would like to propose adding this feature in MRCP V2.=20 =20 Thanks. =20 Jenny=20 =20 --- Jenny Yao Cisco Systems Mail Stop SJ-22/2, 170 W. Tasman Drive, San Jose, CA 95134 Voice: (408)525-3879 ________________________________________________________________ Madhu Gindi Software Development Manager MCEBU ph: 408-525-5571 =09 _____ =20 =09 _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc =09 ------_=_NextPart_001_01C53090.4AB2C4DC Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable
This is assuming the MRCP client will be = generating the=20 different grammar options for approximate. The grammar that needs to be=20 generated for the approximate case is language dependent. As the system = is much=20 of the language intellegience and capbilities is on the server=20 side. 
 
So would the system benefit from=20 an accept=3Dexact/approximate header, that allows the = recogniton/grammar=20 engine on the server to use its language intelligence to = render an=20 equivalent approximate grammar?
 
In implementations such as a VoiceXML = browser that=20 depends on back end MRCP server, there may not be much language=20 intelligence on the server to do this generation for different=20 languages.
 
And if we do decide to provide this header, = the=20 question is where? Regular header in the DEFINE-GRAMMAR = or RECOGNIZE? Or a=20 Content header like Content-Match similar to the Content-Weight that is = being=20 proposed.
I am thinking the later.
 
any thoughts,
Sarvi  

From: speechsc-bounces@ietf.org=20 [mailto:speechsc-bounces@ietf.org] On Behalf Of Dave=20 Burke
Sent: Thursday, March 24, 2005 5:44 AM
To:=20 speechsc@ietf.org; Madhu Gindi
Cc: = jyao@cisco.com
Subject:=20 Re: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using=20 MRCPV2

The "accept" attribute mechanism = described in=20 VoiceXML calls for the generation of a grammar which will = be language=20 and platform dependent.
 
The content of SRGS grammars is = orthogonal to the=20 MRCP protocol and does not belong in it.
 
-- Dave
----- Original Message -----
From:=20 Madhu = Gindi=20
Sent: Thursday, March 24, 2005 = 4:18=20 AM
Subject: [Speechsc] Fwd: = Question for=20 "accept" in VoiceXML 2.0 using MRCP V2

I am forwarding on Jenny's behalf since Jenny's message = bounced=20 back with error. =

/madhu
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
Reply-To: <jyao@cisco.com>
From: = "Jenny Yao=20 \(jyao\)" <jyao@cisco.com>
To:=20 <speechsc@ietf.org>
Cc:=20 "'Sunitha Krishna'" <sunkrish@cisco.com>, = <sarvi@cisco.com>,
  &= nbsp;    =20 "'Madhu Gindi'" <mgindi@cisco.com>
Subject: = Question=20 for "accept" in VoiceXML 2.0 using MRCP V2
Date: Wed, 23 Mar 2005 = 16:36:04 -0800

Our VoiceXML 2.0 = interpreter=20 uses MRCP to communicate with ASR/TTS server. In VoiceXML 2.0, there = is an=20 attribute "accept" for "menu" and "choice" elements. "accept" can be = set to=20 "exact" or "approximate" . If the accept attribute is "exact" then = the user=20 must say the entire phrase in the same order in which they occur in = the=20 choice phrase.If the accept attribute is "approximate", then the = choice may=20 be matched when a user says a subphrase of the expression . We don't = see=20 MRCP V2 draft address this capability. We would like to know if = there is any=20 solution other than adding this to MRCP V2 draft. If not, we would = like to=20 propose adding this feature in MRCP V2.
 
Thanks.
 
Jenny=20
 
---
Jenny Yao
Cisco Systems
Mail Stop SJ-22/2, 170 W. = Tasman Drive,=20 San Jose, CA 95134
Voice: = (408)525-3879


=
________________________________________________________________Madhu=20 Gindi  Software Development Manager MCEBU ph: 408-525-5571=20

_______________________________________________
Speechsc = mailing=20 = list
Speechsc@ietf.org
https://www1.ietf.org/mailman/listinfo/speec= hsc
------_=_NextPart_001_01C53090.4AB2C4DC-- --===============1115288866== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline Content-Transfer-Encoding: 7bit _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc --===============1115288866==-- From speechsc-bounces@ietf.org Thu Mar 24 12:31:50 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA06921 for ; Thu, 24 Mar 2005 12:31:49 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DEWHD-0002vK-46 for speechsc-web-archive@ietf.org; Thu, 24 Mar 2005 12:37:47 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DEWAY-000070-Eg; Thu, 24 Mar 2005 12:30:54 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DEWAX-00006s-Ag for speechsc@megatron.ietf.org; Thu, 24 Mar 2005 12:30:53 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA06796 for ; Thu, 24 Mar 2005 12:30:50 -0500 (EST) Received: from sj-iport-4.cisco.com ([171.68.10.86]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DEWGE-0002tK-WE for speechsc@ietf.org; Thu, 24 Mar 2005 12:36:48 -0500 Received: from sj-core-3.cisco.com (171.68.223.137) by sj-iport-4.cisco.com with ESMTP; 24 Mar 2005 09:30:40 -0800 Received: from vtg-um-e2k1.sj21ad.cisco.com (vtg-um-e2k1.cisco.com [171.70.93.55]) by sj-core-3.cisco.com (8.12.10/8.12.6) with ESMTP id j2OHUPgW016961 for ; Thu, 24 Mar 2005 09:30:36 -0800 (PST) content-class: urn:content-classes:message MIME-Version: 1.0 X-MimeOLE: Produced By Microsoft Exchange V6.0.6603.0 Date: Thu, 24 Mar 2005 09:30:48 -0800 Message-ID: <6677B3346233B94EBB11C060935101200A58D9F4@vtg-um-e2k1.sj21ad.cisco.com> Thread-Topic: Mime content - New headers - Grammar weight Thread-Index: AcUwlzBpveJC0fThS5SAG31fGRNTGg== From: "Shanmugham, Saravanan" To: X-Spam-Score: 0.5 (/) X-Scan-Signature: 7aafa0432175920a4b3e118e16c5cb64 Subject: [Speechsc] Mime content - New headers - Grammar weight X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Content-Type: multipart/mixed; boundary="===============0351794994==" Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.5 (/) X-Scan-Signature: 00e94c813bef7832af255170dca19e36 This is a multi-part message in MIME format. --===============0351794994== content-class: urn:content-classes:message Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C53097.37EE2E59" This is a multi-part message in MIME format. ------_=_NextPart_001_01C53097.37EE2E59 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable =20 =20 I have been looking at RFCs to see if the header proposed by Jeff Kusnitz, Grammar-Weight or a variant could be added to the list of headers that a mime-part can contain. But I can't find it anywhere in the main MIME RFCs if there is an IANA registry or some other mechanism to registering new headers that are permitted in MIME content.=20 =20 Can some body point me in the right direction. =20 Sarvi ------_=_NextPart_001_01C53097.37EE2E59 Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable
 
 
I have = been looking=20 at RFCs to see if the header proposed by Jeff Kusnitz, Grammar-Weight or = a=20 variant could be added to the list of headers that a mime-part can = contain. But=20 I can't find it anywhere in the main MIME RFCs if there is an IANA = registry or=20 some other mechanism to registering new headers that are permitted in = MIME=20 content.
 
Can = some body point=20 me in the right direction.
 
Sarvi
------_=_NextPart_001_01C53097.37EE2E59-- --===============0351794994== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline Content-Transfer-Encoding: 7bit _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc --===============0351794994==-- From speechsc-bounces@ietf.org Thu Mar 24 12:39:06 2005 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA07786 for ; Thu, 24 Mar 2005 12:39:06 -0500 (EST) Received: from megatron.ietf.org ([132.151.6.71]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DEWOF-0003A7-G4 for speechsc-web-archive@ietf.org; Thu, 24 Mar 2005 12:45:04 -0500 Received: from localhost.localdomain ([127.0.0.1] helo=megatron.ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DEWHD-0000z3-AV; Thu, 24 Mar 2005 12:37:47 -0500 Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by megatron.ietf.org with esmtp (Exim 4.32) id 1DEWHB-0000yy-CD for speechsc@megatron.ietf.org; Thu, 24 Mar 2005 12:37:45 -0500 Received: from ietf-mx.ietf.org (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA07692 for ; Thu, 24 Mar 2005 12:37:42 -0500 (EST) Received: from sj-iport-4.cisco.com ([171.68.10.86]) by ietf-mx.ietf.org with esmtp (Exim 4.33) id 1DEWMs-00038Q-BW for speechsc@ietf.org; Thu, 24 Mar 2005 12:43:40 -0500 Received: from sj-core-4.cisco.com (171.68.223.138) by sj-iport-4.cisco.com with ESMTP; 24 Mar 2005 09:37:30 -0800 Received: from vtg-um-e2k1.sj21ad.cisco.com (vtg-um-e2k1.cisco.com [171.70.93.55]) by sj-core-4.cisco.com (8.12.10/8.12.6) with ESMTP id j2OHbQ3U016135; Thu, 24 Mar 2005 09:37:28 -0800 (PST) content-class: urn:content-classes:message MIME-Version: 1.0 X-MimeOLE: Produced By Microsoft Exchange V6.0.6603.0 Subject: RE: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCPV2 Date: Thu, 24 Mar 2005 09:37:34 -0800 Message-ID: <6677B3346233B94EBB11C060935101200A58DA0C@vtg-um-e2k1.sj21ad.cisco.com> Thread-Topic: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCPV2 Thread-Index: AcUwd+MuPnkm6N/7TSiF7nX2YdvwPwAFy3mAAAI3UHA= From: "Shanmugham, Saravanan" To: "Shanmugham, Saravanan" , "Dave Burke" , , "Madhu Gindi" X-Spam-Score: 0.9 (/) X-Scan-Signature: fac892abe0c719c7bb99f6e7c710cdae Cc: jyao@cisco.com X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Content-Type: multipart/mixed; boundary="===============0543875447==" Sender: speechsc-bounces@ietf.org Errors-To: speechsc-bounces@ietf.org X-Spam-Score: 0.9 (/) X-Scan-Signature: a630332fa112280ecc4c4186b5c9ea83 This is a multi-part message in MIME format. --===============0543875447== content-class: urn:content-classes:message Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C53098.29D1643D" This is a multi-part message in MIME format. ------_=_NextPart_001_01C53098.29D1643D Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Or do you see it becoming part of the W3C XML grammar definition. In which case, has this been considered there. =20 =20 Sarvi _____ =20 From: speechsc-bounces@ietf.org [mailto:speechsc-bounces@ietf.org] On Behalf Of Shanmugham, Saravanan Sent: Thursday, March 24, 2005 8:41 AM To: Dave Burke; speechsc@ietf.org; Madhu Gindi Cc: jyao@cisco.com Subject: RE: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCPV2 This is assuming the MRCP client will be generating the different grammar options for approximate. The grammar that needs to be generated for the approximate case is language dependent. As the system is much of the language intellegience and capbilities is on the server side.=20 =20 So would the system benefit from an accept=3Dexact/approximate header, that allows the recogniton/grammar engine on the server to use its language intelligence to render an equivalent approximate grammar? =20 In implementations such as a VoiceXML browser that depends on back end MRCP server, there may not be much language intelligence on the server to do this generation for different languages. =20 And if we do decide to provide this header, the question is where? Regular header in the DEFINE-GRAMMAR or RECOGNIZE? Or a Content header like Content-Match similar to the Content-Weight that is being proposed. I am thinking the later. =20 any thoughts, Sarvi =20 _____ =20 From: speechsc-bounces@ietf.org [mailto:speechsc-bounces@ietf.org] On Behalf Of Dave Burke Sent: Thursday, March 24, 2005 5:44 AM To: speechsc@ietf.org; Madhu Gindi Cc: jyao@cisco.com Subject: Re: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCPV2 The "accept" attribute mechanism described in VoiceXML calls for the generation of a grammar which will be language and platform dependent.=20 =20 The content of SRGS grammars is orthogonal to the MRCP protocol and does not belong in it. =20 -- Dave ----- Original Message -----=20 From: Madhu Gindi =20 To: speechsc@ietf.org=20 Cc: jyao@cisco.com=20 Sent: Thursday, March 24, 2005 4:18 AM Subject: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using MRCP V2 I am forwarding on Jenny's behalf since Jenny's message bounced back with error.=20 =09 /madhu = =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D =09 Reply-To: From: "Jenny Yao \(jyao\)" To: Cc: "'Sunitha Krishna'" , , "'Madhu Gindi'" Subject: Question for "accept" in VoiceXML 2.0 using MRCP V2 Date: Wed, 23 Mar 2005 16:36:04 -0800 =09 Our VoiceXML 2.0 interpreter uses MRCP to communicate with ASR/TTS server. In VoiceXML 2.0, there is an attribute "accept" for "menu" and "choice" elements. "accept" can be set to "exact" or "approximate" . If the accept attribute is "exact" then the user must say the entire phrase in the same order in which they occur in the choice phrase.If the accept attribute is "approximate", then the choice may be matched when a user says a subphrase of the expression . We don't see MRCP V2 draft address this capability. We would like to know if there is any solution other than adding this to MRCP V2 draft. If not, we would like to propose adding this feature in MRCP V2.=20 =20 Thanks. =20 Jenny=20 =20 --- Jenny Yao Cisco Systems Mail Stop SJ-22/2, 170 W. Tasman Drive, San Jose, CA 95134 Voice: (408)525-3879 ________________________________________________________________ Madhu Gindi Software Development Manager MCEBU ph: 408-525-5571 =09 _____ =20 =09 _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc =09 ------_=_NextPart_001_01C53098.29D1643D Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable
Or do you see it becoming part of the W3C = XML grammar=20 definition. In which case, has this been considered = there.
 
 
Sarvi

From: speechsc-bounces@ietf.org=20 [mailto:speechsc-bounces@ietf.org] On Behalf Of Shanmugham,=20 Saravanan
Sent: Thursday, March 24, 2005 8:41 AM
To: = Dave=20 Burke; speechsc@ietf.org; Madhu Gindi
Cc:=20 jyao@cisco.com
Subject: RE: [Speechsc] Fwd: Question for = "accept" in=20 VoiceXML 2.0 using MRCPV2

This is assuming the MRCP client will be = generating the=20 different grammar options for approximate. The grammar that needs to be=20 generated for the approximate case is language dependent. As the system = is much=20 of the language intellegience and capbilities is on the server=20 side. 
 
So would the system benefit from=20 an accept=3Dexact/approximate header, that allows the = recogniton/grammar=20 engine on the server to use its language intelligence to = render an=20 equivalent approximate grammar?
 
In implementations such as a VoiceXML = browser that=20 depends on back end MRCP server, there may not be much language=20 intelligence on the server to do this generation for different=20 languages.
 
And if we do decide to provide this header, = the=20 question is where? Regular header in the DEFINE-GRAMMAR = or RECOGNIZE? Or a=20 Content header like Content-Match similar to the Content-Weight that is = being=20 proposed.
I am thinking the later.
 
any thoughts,
Sarvi  

From: speechsc-bounces@ietf.org=20 [mailto:speechsc-bounces@ietf.org] On Behalf Of Dave=20 Burke
Sent: Thursday, March 24, 2005 5:44 AM
To:=20 speechsc@ietf.org; Madhu Gindi
Cc: = jyao@cisco.com
Subject:=20 Re: [Speechsc] Fwd: Question for "accept" in VoiceXML 2.0 using=20 MRCPV2

The "accept" attribute mechanism = described in=20 VoiceXML calls for the generation of a grammar which will = be language=20 and platform dependent.
 
The content of SRGS grammars is = orthogonal to the=20 MRCP protocol and does not belong in it.
 
-- Dave
----- Original Message -----
From:=20 Madhu = Gindi=20
Sent: Thursday, March 24, 2005 = 4:18=20 AM
Subject: [Speechsc] Fwd: = Question for=20 "accept" in VoiceXML 2.0 using MRCP V2

I am forwarding on Jenny's behalf since Jenny's message = bounced=20 back with error. =

/madhu
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
Reply-To: <jyao@cisco.com>
From: = "Jenny Yao=20 \(jyao\)" <jyao@cisco.com>
To:=20 <speechsc@ietf.org>
Cc:=20 "'Sunitha Krishna'" <sunkrish@cisco.com>, = <sarvi@cisco.com>,
  &= nbsp;    =20 "'Madhu Gindi'" <mgindi@cisco.com>
Subject: = Question=20 for "accept" in VoiceXML 2.0 using MRCP V2
Date: Wed, 23 Mar 2005 = 16:36:04 -0800

Our VoiceXML 2.0 = interpreter=20 uses MRCP to communicate with ASR/TTS server. In VoiceXML 2.0, there = is an=20 attribute "accept" for "menu" and "choice" elements. "accept" can be = set to=20 "exact" or "approximate" . If the accept attribute is "exact" then = the user=20 must say the entire phrase in the same order in which they occur in = the=20 choice phrase.If the accept attribute is "approximate", then the = choice may=20 be matched when a user says a subphrase of the expression . We don't = see=20 MRCP V2 draft address this capability. We would like to know if = there is any=20 solution other than adding this to MRCP V2 draft. If not, we would = like to=20 propose adding this feature in MRCP V2.
 
Thanks.
 
Jenny=20
 
---
Jenny Yao
Cisco Systems
Mail Stop SJ-22/2, 170 W. = Tasman Drive,=20 San Jose, CA 95134
Voice: = (408)525-3879


=
________________________________________________________________Madhu=20 Gindi  Software Development Manager MCEBU ph: 408-525-5571=20

_______________________________________________
Speechsc = mailing=20 = list
Speechsc@ietf.org
https://www1.ietf.org/mailman/listinfo/speec= hsc
------_=_NextPart_001_01C53098.29D1643D-- --===============0543875447== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline Content-Transfer-Encoding: 7bit _______________________________________________ Speechsc mailing list Speechsc@ietf.org https://www1.ietf.org/mailman/listinfo/speechsc --===============0543875447==--