Return-Path: X-Original-To: speechsc@core3.amsl.com Delivered-To: speechsc@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 43C943A67E4 for ; Tue, 14 Jul 2009 02:44:24 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -1.202 X-Spam-Level: X-Spam-Status: No, score=-1.202 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HTML_MESSAGE=0.001, MIME_QP_LONG_LINE=1.396] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id bF13SQUTaD9K for ; Tue, 14 Jul 2009 02:44:22 -0700 (PDT) Received: from n76.bullet.mail.sp1.yahoo.com (n76.bullet.mail.sp1.yahoo.com [98.136.44.48]) by core3.amsl.com (Postfix) with SMTP id DF6E53A68F0 for ; Tue, 14 Jul 2009 02:44:22 -0700 (PDT) Received: from [216.252.122.219] by n76.bullet.mail.sp1.yahoo.com with NNFMP; 14 Jul 2009 09:44:25 -0000 Received: from [67.195.9.82] by t4.bullet.sp1.yahoo.com with NNFMP; 14 Jul 2009 09:44:24 -0000 Received: from [67.195.9.98] by t2.bullet.mail.gq1.yahoo.com with NNFMP; 14 Jul 2009 09:44:24 -0000 Received: from [127.0.0.1] by omp102.mail.gq1.yahoo.com with NNFMP; 14 Jul 2009 09:44:24 -0000 X-Yahoo-Newman-Property: ymail-3 X-Yahoo-Newman-Id: 704360.22074.bm@omp102.mail.gq1.yahoo.com Received: (qmail 7838 invoked by uid 60001); 14 Jul 2009 09:44:24 -0000 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=yahoo.com; s=s1024; t=1247564664; bh=6PGtjmTXhy7iihojsv08oW5n0ww+/Jh8Xuq5Hpgw1Ds=; h=Message-ID:X-YMail-OSG:Received:X-Mailer:References:Date:From:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type; b=kjrP4kcJsGw02o2JbkYiJ0TtUHFGQsEkPiRM2ExIzl2kbXMnYm3fUSk1htmyHWoLHFi7VamT8nOD8VNYb2n0tRo97cBIo0JLdpNZoJqlYFOr1dXXoFYkGAUqkX+0GZ0GkJBUiYhcUFsvTIwKVFwHOfSxFRo8EdGw3ZJ8slaWRt4= DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:X-YMail-OSG:Received:X-Mailer:References:Date:From:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type; b=SSqhwA+BmdddtSmqOh9YtgAPoNjq/QKounPRHLADvW3Bwi9q6o/8BJ9HmJEwLXA5DW+ZaSZAIWTyWwTcKG6BcUac/qXvuXw6jVmlDKJsNjFtijFcqbxP79d91GYhvvP6PrIetpmpIa59C3+ikzcfjJHuI+nFJP77ZAMGdcFo160=; Message-ID: <441543.7572.qm@web111316.mail.gq1.yahoo.com> X-YMail-OSG: eM1wf8kVM1kaMWG0pcbN15Qa8D7mOp9TThBq1rX7mzGowibXVY.wTuOv7tmqRATaOj4f.KoL2stoj0fcg6tWPerSYc2an0xbuj0LzHrcjfF7x7zYrLXwMWRfRjoJECXT7qpPgvBp3L.zCFCIsbmFAS_AcTcz.72DSori4qT1XXjmw6rbjzPOMCzlUd1IRLroY3YRP285MXlrSKmpq2vMSR8.I2BRCboRDjdsJYyGtVOBWyo9ziJ.3CpBrhl3sI6fgE80khvFKg9fLaByPaT7sohMiHdWWFZXnLkQMn.4B_MBnfEnxggFD3QlypCj1uR6wALZ0N9tWAyt08TwsTFO7nheLMrRGv5AywZjg_gwYZh7y9_VQa68sg-- Received: from [91.198.247.201] by web111316.mail.gq1.yahoo.com via HTTP; Tue, 14 Jul 2009 02:44:23 PDT X-Mailer: YahooMailRC/1358.21 YahooMailWebService/0.7.289.15 References: <033101c9ff3a$cbe33160$63a99420$%roni@huawei.com> Date: Tue, 14 Jul 2009 02:44:23 -0700 (PDT) From: Arsen Chaloyan To: Roni Even , sarvi@cisco.com, dburnett@voxeo.com In-Reply-To: <033101c9ff3a$cbe33160$63a99420$%roni@huawei.com> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="0-1520693546-1247564663=:7572" Cc: speechsc@ietf.org, oran@cisco.com, rai@ietf.org Subject: Re: [Speechsc] RAI review of draft-ietf-speechsc-mrcpv2-19 X-BeenThere: speechsc@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: Speech Services Control Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 14 Jul 2009 09:44:24 -0000 --0-1520693546-1247564663=:7572 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Hi,=0ATwo small issues left you may want to fix.=0A=0A- Section 9.4.30=0AMe= dia type should be "text/plain" instead of "plain/text"=0Ahttp://www.w3.org= /TR/2004/REC-speech-synthesis-20040907/=0A=0A- Find BARGE-IN-OCCURED replac= e with BARGE-IN-OCCURRED=0AThere are 3 such occurrences in the text.=0A=0AT= hanks,=0AArsen.=0A=0A=0A=0A=0A________________________________=0AFrom: Roni= Even =0ATo: sarvi@cisco.com; dburnett@voxeo.com=0ACc= : speechsc@ietf.org; oran@cisco.com; rai@ietf.org=0ASent: Wednesday, July 8= , 2009 12:40:17 AM=0ASubject: [Speechsc] RAI review of draft-ietf-speechsc-= mrcpv2-19=0A=0A =0AHi,=0AI was=0Aassigned to do a RAI review of the draft. = The draft looks ready for=0Apublication to me. I have some comments mostly= editorial. =0AThe=0Aonly issue I see that is not pure editorial is the iss= ue of the different=0Aparameters like confidence threshold, sensitivity lev= el (see comments 11, 13,=0A15, 16 and 17). I think that some clarification = on the semantics and the scale=0A(for example are the values linearly space= d) as well as when they are useful=0Awill be helpful to implementers.=0A1. = In figure 1 Expand the abbreviations TTS, ASR, SV , SI=0Aand how they= are related to the media resource types in 3.1=0A2. In figure 1 ther= e is a SIP dialog between the MRCPv2=0Aclient and the media source/sink, wh= at is this dialog, I only saw in section 4=0Aa dialog between the client an= d server.=0A3. In section 3.2 you have =E2=80=9CFor=0Aexample: sip:mr= cpv2@example.net=E2=80=9D twice one after=0Athe other.=0A =0A4. In th= e example in section 4.2 you=0A=E2=80=9Ca=3Dcmid:1=E2=80=9D, cmid is specif= ied later in the document so maybe you=0Acan add some reference to where it= is specified=0A =0A5. In the example is section 4.2 and in=0Afollowi= ng examples you have =E2=80=9Cm=3Daudio 49170 RTP/AVP 0 96=E2=80=9D but do = not=0Ahave an rtpmap parameter for mapping 96 (dynamic payload type number)= to a=0Amedia encoding name.=0A =0A6. In section 4.3 =E2=80=9CAlso no= te that=0Amore that one media session can be associated with a single resou= rce if need=0Abe, but this scenario is not useful for the current set of re= sources=E2=80=9D.=0AThere is a typo the second =E2=80=9Cthat=E2=80=9D shoul= d be =E2=80=9Cthan=E2=80=9D. I=0Aam also not sure if the current syntax in = this document can support the mode.=0A =0A =0A7. In section 4.3 =E2= =80=9CThe formatting=0Aof the"cmid" attribute in SDP RFC3388 [RFC4566]=E2= =80=9D. I think you=0Ameant SDP grouping and need the reference to RFC 3388= ..=0A =0A =0A8. In section 5.1 =E2=80=9CThe message-length=0Afield spe= cifies the length of the message, including the start-line=E2=80=9D is=0Ath= e length in Bytes, there is no unit specified.=0A =0A9. In section 6.= 3.1, typo you have=0A=E2=80=9CVerfication =E2=80=9C instead of verification= .. It appears twice in the=0Asection.=0A =0A10. In the example in section = 7 you have=0A=E2=80=9Cm=3Daudio 0 RTP/AVP 0 1 3=E2=80=9D payload type 1 was= deleted from the IANA=0Aregistry, maybe have another payload type number.= =0A =0A11. In section 9.4.1, 9.4.2 and 9.4.3=0Ayou specify confidence thr= eshold, sensitivity level and speed vs accuracy. What=0Ais the scale here; = is it linear between 0 and 1. What is the absolute value of=0Athe number, i= f you receive the same confidence level from two recognizers are=0Athey the= same (e.g. when using context block to switch servers). For the=0Aspeed v= s accuracy, how does the client know what is the relation between the=0Aval= ue and the number of available sessions, since this seems to be the reason= =0Afor using this parameter.=0A =0A12. In 9.4.9 and in 10.4.8, 11.4.11 wh= at=0Aare the values for media-type-value, you also mention audio and video = but it=0Alooks to me that this document only discusses voice.=0A =0A13. I= n 9.4.35 and 9.4.36 what is the=0Ascale for the consistency here. How does = one know what close means. What is the=0Aconsistency between different reco= gnizers.=0A =0A14. In section 9.6.3.3 in the example=0A(figure 2) confide= nce should be 0.75 and not 75=0A =0A15. In section 10.4.1 it is not clear= =0Ahow you measure the sensitivity in order to specify, is it based on some= SNR=0Atranslated to 0 to 1 scale?=0A =0A16. In 11.4.6 the same issue wit= h the=0Ascale, how does the client know how to set a value when working wit= h different=0Aspeaker verification servers.=0A =0A17. In 11.5.2.9 you sta= te that the=0Averification-score is not a probability, so what is it. How c= an the client decide=0Aif, for example, 0 is a good score for specifying th= e threshold. I also=0Anoticed that the values in the example in section 11= ..5.2.10 are very precise=0Alike 0.98514 is this the expected precision. The= examples here and in section=0A11.11 do not show the threshold, if the thr= eshold is required for this flow why=0Anot show it in the example?=0A =0A18= .. In section 12.3 the suggestion is to=0Ause SRTP as the mandatory intero= perability mode. If the reason for mandating=0ASRTP is for a common mode yo= u should also decide on a key exchange mechanism. I=0Asuggest you look at h= ttp://tools.ietf.org/html/draft-ietf-avt-srtp-not-mandatory-02 for discussi= on on media security.=0A =0A19. In section13.7.2 you specify the attribut= e=0Aresource as session level yet in the example in section 4.2 it is a med= ia level=0Aattribute. The same goes for the channel attribute=0A =0AThanks= =0A =0ARoni=0AEven --0-1520693546-1247564663=:7572 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: quoted-printable
Hi,
Two small issues left you may want to fix.

- Section = 9.4.30
Media type should be "text/plain" instead of "plain/text"
http://www.w3.org/TR/2004/REC-speech-synthesis-20040907/<= /span>

- Find BARGE-IN-OCCURED replace with BARGE-IN-OCCURRED
The= re are 3 such occurrences in the text.

Thanks,
Arsen.

=
From: Roni Even <Even.roni@huawei.com><= br>To: sarvi@cisco.com; dburnett@voxeo.com
Cc:= speechsc@ietf.org; oran@cisco.com; rai@ietf.org
Sent: Wednesday, July 8, 2009 12:40:17 AM
<= span style=3D"font-weight: bold;">Subject: [Speechsc] RAI review= of draft-ietf-speechsc-mrcpv2-19

=0A=0A=0A=0A =0A =0A=0A=0A=0A=0A
=0A=0A

I was=0Aassigned to do a RAI review of the draf= t.  The draft looks ready for=0Apublication to me. I have some comment= s mostly editorial.

=0A=0A

The=0Aonly issue I see that is n= ot pure editorial is the issue of the different=0Aparameters like confidenc= e threshold, sensitivity level (see comments 11, 13,=0A15, 16 and 17). I th= ink that some clarification on the semantics and the scale=0A(for example a= re the values linearly spaced) as well as when they are useful=0Awill be he= lpful to implementers.

=0A=0A

1.      =0A= In figure 1 Expand the abbreviations TTS, ASR, SV , SI=0Aand how t= hey are related to the media resource types in 3.1

=0A=0A

2.    &nb= sp; =0AIn figure 1 there is a SIP dialog betwee= n the MRCPv2=0Aclient and the media source/sink, what is this dialog, I onl= y saw in section 4=0Aa dialog between the client and server.

=0A= =0A

<= span style=3D"">3.      =0AIn section 3.2 you have =E2=80= =9CFor=0Aexample: sip:mrcpv2@example.net=E2=80=9D twice one a= fter=0Athe other.

=0A=0A

 

=0A=0A

4.     &nb= sp;=0AIn the e= xample in section 4.2 you=0A=E2=80=9Ca=3Dcmid:1=E2=80=9D, cmid is specified= later in the document so maybe you=0Acan add some reference to where it is= specified

=0A=0A

 = ;

=0A=0A

= 5.      =0A<= /span>In the example = is section 4.2 and in=0Afollowing examples you have =E2=80=9Cm=3Daudio 4917= 0 RTP/AVP 0 96=E2=80=9D but do not=0Ahave an rtpmap parameter for mapping 9= 6 (dynamic payload type number) to a=0Amedia encoding name.

=0A= =0A

 

=0A=0A

6.      =0AIn section 4.3 =E2=80=9CAlso note that= =0Amore that one media session can be associated with a single resource if = need=0Abe, but this scenario is not useful for the current set of resources= =E2=80=9D.=0AThere is a typo the second =E2=80=9Cthat=E2=80=9D should be = =E2=80=9Cthan=E2=80=9D. I=0Aam also not sure if the current syntax in this = document can support the mode.

=0A=0A

<= span style=3D"font-size: 11pt; font-family: "Calibri","sans-= serif";">  

=0A=0A

 

=0A=0A

7.     = ; =0AIn s= ection 4.3 =E2=80=9CThe formatting=0Aof the"cmid" attribute in SDP RFC3388 = [RFC4566]=E2=80=9D. I think you=0Ameant SDP grouping and need the reference= to RFC 3388.

=0A=0A

&n= bsp;

=0A=0A

 

=0A=0A

8.      =0A= In section 5.1 =E2=80= =9CThe message-length=0Afield specifies the length of= the message, including the start-line=E2=80=9D is=0Athe length in Bytes, t= here is no unit specified.

=0A=0A

 

=0A=0A

9.    =   =0AIn section 6.3.1, typo you have=0A=E2=80=9CVerfication =E2=80=9C instead o= f verification. It appears twice in the=0Asection.

=0A=0A

 

=0A=0A

10.=   =0AIn the example in section 7 you have=0A=E2=80=9Cm=3Daudio 0 RTP/AVP 0 1= 3=E2=80=9D payload type 1 was deleted from the IANA=0Aregistry, maybe have= another payload type number.

=0A=0A

 

=0A=0A

11.  =0AIn section 9.4.1, 9= ..4.2 and 9.4.3=0Ayou specify confidence threshold, sensitivity level and sp= eed vs accuracy. What=0Ais the scale here; is it linear between 0 and 1. Wh= at is the absolute value of=0Athe number, if you receive the same confidenc= e level from two recognizers are=0Athey the same (e.g. when using context b= lock to switch servers).  For the=0Aspeed vs accuracy, how does the cl= ient know what is the relation between the=0Avalue and the number of availa= ble sessions, since this seems to be the reason=0Afor using this parameter.=

=0A=0A

  =0A=0A

12.  =0AIn 9.4.9 and in 10.4.8, 11.4.11 what=0Aare the v= alues for media-type-value, you also mention audio and video but it=0Alooks= to me that this document only discusses voice.

=0A=0A

 

=0A=0A

13.<= span style=3D"font-family: "Times New Roman"; font-style: normal;= font-variant: normal; font-weight: normal; font-size: 7pt; line-height: no= rmal; font-size-adjust: none; font-stretch: normal; -x-system-font: none;">=   =0AIn 9.4.35 and 9.4.36 what is the=0Ascale for the consistency here. How doe= s one know what close means. What is the=0Aconsistency between different re= cognizers.

=0A=0A

 = ;

=0A=0A

= 14.  =0AIn section 9.6.3.3 in the example=0A(f= igure 2) confidence should be 0.75 and not 75

=0A=0A

 

=0A=0A

15.&nbs= p; =0AIn = section 10.4.1 it is not clear=0Ahow you measure the sensitivity in order t= o specify, is it based on some SNR=0Atranslated to 0 to 1 scale?

= =0A=0A

 

=0A=0A16.  =0AIn 11.4.6 the same issue with the=0Ascale, how does the cli= ent know how to set a value when working with different=0Aspeaker verificat= ion servers.

=0A=0A

&nb= sp;

=0A=0A

17.  =0A<= span dir=3D"ltr">In 11.5.2.9 you state that the=0Aver= ification-score is not a probability, so what is it. How can the client dec= ide=0Aif, for example, 0 is a good score for specifying the threshold. = ; I also=0Anoticed that the values in the example in section 11.5.2.10 are = very precise=0Alike 0.98514 is this the expected precision. The examples he= re and in section=0A11.11 do not show the threshold, if the threshold is re= quired for this flow why=0Anot show it in the example?

=0A=0A

 

=0A=0A

18.  =0AIn section 12.3 the suggestion is to=0Ause SRTP as the mandator= y interoperability mode. If the reason for mandating=0ASRTP is for a common= mode you should also decide on a key exchange mechanism. I=0Asuggest you l= ook at http://tools.ietf.org/html/draft-ietf-avt-srtp-no= t-mandatory-02=0Afor discussion on media security.

= =0A=0A

 

=0A=0A

19.  =0A= In section13.7.2 you specify the attribute=0Aresource as ses= sion level yet in the example in section 4.2 it is a media level=0Aattribut= e. The same goes for the channel attribute

=0A=0A

 

=0A=0A

T= hanks

=0A=0A

 

=0A=0A

Roni=0AEven =0A=0A

 

=0A=0A=

 

=0A=0A
=0A=0A
--0-1520693546-1247564663=:7572--