Re: Comments about SSML from ITS WG

Hello Dan, all ,

this is a private reply.

Dan Burnett さんは書きました:
>
> Dear Jirka (and ITS WG),
>
> Thank you for your comments. Our responses are embedded below, 
> preceded by "DB>>".
> If you have any concerns with our responses, please let us know. If we 
> do not hear from you within two weeks of today we will assume that you 
> have accepted our resolutions.
>
> Dan Burnett
> SSML 1.1 Co-Editor
> Voice Browser Working Group
>
> On Jul 17, 2008, at 9:50 AM, Jirka Kosek wrote:
>> Hello,
>>
>> ITS WG reviewed SSML 1.1 
>> (http://www.w3.org/TR/2008/WD-speech-synthesis11-20080620/) from 
>> perspective of suitability for internationalization and localization.
>>
>> We would appreciate if you can incorporate our feedback into SSML.
>>
>> SSML should allow usage of ITS markup (http://www.w3.org/TR/its/), 
>> because SSML documents contain natural text with speech markup and as 
>> such it is very likely that they can be translated.
>>
>> SSML currently doesn't provide direct support for ITS and has limited 
>> extensibility which prevents using ITS (see below).
>>
>> Problem 1:
>> ==========
>>
>> Section 2.2.3 
>> (http://www.w3.org/TR/2008/WD-speech-synthesis11-20080620/#S2.2.3) 
>> allows usage of foreign elements inside SSML documents. Thus it is 
>> legitimate to use elements like its:rules or its:ruby in SSML documents.
>>
>> However specification doesn't explicitly allow to use foreign 
>> attributes. This means that it is not possible to use local ITS 
>> attributes like its:translate.
>>
>> Solution 1:
>> ===========
>>
>> SSML should allow foreign attributes on any SSML element. Ideally 
>> such possibility will not be mentioned only in prose of spec, but 
>> also XML schema will specify this using xs:anyAttribute (and 
>> similarly by xs:any for elements).
>
> DB>> Resolution: Accepted with modifications
> DB>> Explanation: We agree that non-SSML attributes should be 
> permitted in the same way that non
> DB>> SSML elements are permitted, and in fact the other subsections of 
> 2.2 describe this in detail. We
> DB>> will extend the text in 2.2.3 to indicate that attributes are 
> also allowed. However, your request to
> DB>> allow ITS attributes does not, in our opinion, introduce any more 
> need to generalize the Schema
> DB>> than was already present in SSML 1.0 (and hence 1.1). We believe 
> that converting the Schema
> DB>> to use xs:anyAttribute and xs:any would trivialize the Schema. We 
> do not plan to change the
> DB>> Schema.


My reaction may not come to you as a surprise ... I would propose a 
similar resolution as for the PLS 1.0 schema. Both in PLS 1.0 and SSML 
1.1 you already have one element for general, additional markup. In SSML 
1.1 this is the metadata element which has both element and attribute 
extensibility. I understand that you do not want to introduce general 
element extensibility, but I would propose you to do the same as I 
proposed for PLS 1.0: add general attribute extensibility to the SML 
1.1. schema. I think this is harmless and a huge benefit for "foreign 
vocabularies" like ITS.

>
>>
>> Problem 2:
>> ==========
>>
>> Element sub have attribute alias 
>> (http://www.w3.org/TR/2008/WD-speech-synthesis11-20080620/#S3.1.11) 
>> which contains text for pronuncation. However it is not possible to 
>> attach any ITS category to single attribute. Moreover using 
>> attributes for natural language text is against XML I18N BP 
>> (http://www.w3.org/TR/2008/NOTE-xml-i18n-bp-20080213/#DevAttributes). 
>> Using subelement for alias will be better solution.
>>
>> Solution 2:
>> ===========
>>
>> Use alias subelement instead of attribute.
>>
>
> DB>> Resolution: Rejected (deferred)
> DB>> Explanation: We agree that the inability to add markup to the 
> spoken text of the <sub> element
> DB>> is a lack within SSML. This comment was raised during the Last 
> Call Working Draft for SSML 1.0
> DB>> (see SSCR145-46 at
> DB>> 
> http://www.w3.org/TR/2003/CR-speech-synthesis-20031218/disposition.html#SSCR145-46). 
>
> DB>> We still believe that a change of this sort will break 
> compatibility with SSML 1.0, and that the
> DB>> stated scope of SSML 1.1 (see Section 1.2 in
> DB>> http://www.w3.org/TR/2007/WD-ssml11reqs-20070611/#intro) does not 
> suggest making this sort
> DB>> of change. We propose to defer this change, as before, to a 
> future version where SSML is more
> DB>> broadly re-written.

I personally understand and agree with your position to keep the alias 
attribute for backwards compatibilty reasons.

Felix

>
>> Regards,
>>
>> Jirka Kosek
>> on behalf of ITS WG
>>
>> -- 
>> ------------------------------------------------------------------
>> Jirka Kosek e-mail: jirka@kosek.cz http://xmlguru.cz
>> ------------------------------------------------------------------
>> Professional XML consulting and training services
>> DocBook customization, custom XSLT/XSL-FO document processing
>> ------------------------------------------------------------------
>> OASIS DocBook TC member, W3C Invited Expert, ISO JTC1/SC34 member
>> ------------------------------------------------------------------
>>
>
>

Received on Thursday, 18 September 2008 03:18:20 UTC