SSML 1.1 April face-to-face meeting summary

The SSML subgroup of the Voice Browser Working Group held its ninth  
meeting in Beijing on April 22-24, 2008.  The meeting included  
attendees by IBM, iFlyTek, France Telecom, Toshiba, and Voxeo.  The  
China office of W3C was kind enough to host us at the same time they  
were hosting WWW2008!


The primary goals of this meeting were to review and address any  
outstanding change requests, review and address any comments from  
public reviewers of the fifth Working Draft, to make progress on the  
Pronunciation Alphabet Registry, and to prepare for the Last Call  
Working Draft.

We made huge progress on two topics:  a logarithmic control for the  
"volume" attribute of <prosody> and an informal Internationalization  
group review of the specification.  A member of the SYMM working  
group (the producers of SMIL) joined us to review our proposal for  
logarithmic volume control.  As a result, we were asked to submit a  
proposal to the SYMM WG to change the "soundLevel" attribute of SMIL  
to use this logarithmic scale.  With a small amount of additional  
discussion, the topic of logarithmic control of volume and sound  
level in SSML should be completed before the next SSML Working Draft.

A member of the W3C Internationalization Working Group joined us to  
review the changes in the specification over the past year.  The most  
significant outcome was the realization that the specification does  
not clearly cross-link the different language- and voice-controlling  
aspects of the document (xml:lang, onlangfailure, onvoicefailure),  
does not explain how the languages supported by a voice are  
described, and does not explain what happens when the voice selection  
algorithm fails to find any matching voice.  These problems will be  
fixed in the next draft of SSML 1.1.

As a group we randomly reviewed implementation report tests written  
by different people, just to verify that the tests were being written  
properly.  This resulted in some good discussion and, ultimately, a  
change in how a number of tests were written to reduce the amount of  
customization that would be required of implementers.

Finally, we continued discussion on the Pronunciation Alphabet  
Registry.   We will note in the specification that the official  
location of the registry can be found on the SSML namespace page.  We  
also discussed potential security considerations of the registry in  
preparation for converting the document to be an IETF draft.

We continue to have weekly teleconferences to work on issues that  
arise and to track progress on the remaining IR assertions and tests.

We expect to publish a Last Call Working Draft of the specification  
before our next meeting.

Our next meeting will be held in Lanzhou, China, from July 22-24.


Daniel C. Burnett and 双志伟 (Zhi Wei Shuang)
Co-chairs, Speech Synthesis Subgroup, VBWG

Received on Monday, 12 May 2008 17:02:38 UTC