W3C

– DRAFT –
Spoken Pronunciation Task Force Teleconference

23 June 2021

Attendees

Present
Irfan, JF, mhakkinen, paul_grenier, SteveNoble_
Regrets
-
Chair
Paul_Grenier
Scribe
Irfan

Meeting minutes

Agenda Review & Announcements

mhakkinen: we have submitted the abstract in ATIA. second round kicks off later in the summer

Review Feedback

https://github.com/w3c/pronunciation/issues

https://github.com/w3c/pronunciation/issues/90

mhakkinen: there is another ticket about CSS.

mhakkinen: biggest gap in CSS approach is that it doesnt provide all the solutions

paul_grenier: Based on our gap analysis document, CSS speech has no mechanism for language, phonetic pronunciation, substitution, and "say as" (feature).

https://github.com/w3c/pronunciation/issues/84

JF: I am concerned about i18 problem. are screen reader actually doing the translation?

mhakkinen: its not SR but synthesizer

JF: have we thought about attribute and value where we can enforce SR reader to read it correctly.

JF: I do not think is it really a problem that we should be fixing.

JF: SR is fixing that problem

janina: dd:mm:yyyy is an example where we may need some test cases

janina: locale settings should help for the issue entered in #84. we can't fix everything

paul_grenier: its up to author of TTS to figure out certain things. this is not for us to solve. we can't fix bad authoring

paul_grenier: we have a date method in our document where we can provide some examples.

paul_grenier: there is a format attribute as date.

JF: looking at working group note where it got that format attribute which can help answering COGA question.

<JF> https://www.w3.org/TR/ssml-sayas/#S3.1

JF: Doesn't solve the authoring problem but there is a mechanism.

janina: we need to draft a response and send it to COGA as a formal response.

https://github.com/w3c/pronunciation/issues/85

<JF> Google's Media 'tag': https://developers.google.com/assistant/conversational/ssml#media

JF: is there anything actionable here?

paul_grenier: do we want to incorporate some of these notes into our documents?

janina: We are not ready to respond on this yet. Let us see other comments and arguments as well.

JF: we can acknowledge the comments.

https://github.com/w3c/pronunciation/issues/86

mhakkinen: issue is here that one side is not using a11y API but DOM. he is right his approach that its going to be more expensive for him.

paul_grenier: even without the a11y API its using data set programmatically that would provide lot of speed than query selector.

mhakkinen: we are not aiming not to use any API.

paul_grenier: will make a note to myself to do some test with treewalker and data set.

https://github.com/w3c/pronunciation/issues/87

mhakkinen: may be we can have something like aria-ssml or something

mhakkinen: have questions about outreach. contacted MS but heard nothing. Also haven't heard from NVDA. Those are the implementors that we need to hear from

Minutes manually created (not a transcript), formatted by scribe.perl version 136 (Thu May 27 13:50:24 2021 UTC).

Diagnostics

Maybe present: janina