Meeting minutes
matatk did that work?
Our meeting link https://
<Irfan> https://
<matatk> Meeting notes: https://
<matatk> Also: https://
matatk: we're basically in the same place we were...
… we're at an impass with ARIA because the 'large platforms' do not see a path to implementing all of our requirements
… Paul offered another approach to possibly help smooth the path forward with them but the more concrete stuff we got from other meetings included a lot of interest from WHATWG folks who had never heard of this work before
… they also asked if we could use CSS but CSSWG responded with 'no, it's not appropriate'
Agenda Review, Membership & Announcements
matatk: WHATWG asked about using Ruby but we think they misunderstood that we don't need visual rendering in all cases
… Simon Peters suggested separate attributes because the single-attribute authoring wasn't seen as ideal
… SSML most likely will not get into HTML as a first-class object (ala SVG) per WHATWG
… so the multi-attribute mapped to a single "place" in the axTree is still a possibility
PaulG: The idea was to allow the AT to delegate pronunciation.
… This could be done via the Web Speech API in the UA.
… This gives them an opt-out. One thing that came up in our call last week was that some UAs need to be online to generate speech (even if they support SSML).
… This doesn't work for our use cases, some of which need offline access.
TPAC
Sarah: What was the feature that was casuing them the most problems?
PaulG: Pausing, breaks.
Irfan: Some platforms thought it would be difficult for them to implement (possibly at all), but others do use SSML in some places, and seemed like they may be able to implement some of it for pronunication, perhpas.
PaulG: Even with author-controlled pronunication [proposed above], AT users may not opt in becuase it won't match their chosen TTS preferences (e.g. speed).
… Maybe puasing/breaks are causing them issues in relation to settings like this.
… Would love to know what the implementation headaches they forsee are.
Irfan: Talkinkg of the business case, we _do_ have a clear business case for the critical use cases we identified.
… Can we send a letter about those business cases?
… We've tried to keep them updated about our progress, and have limited to 5 of our use cases.
PaulG: Maybe we need an example for pause; can we provide one?
Alan: I can provide some; we use breaks a lot.
PaulG: The use case we have doesn't perhaps seem clear.
Alan: One thing that comes to mind is if you have tables, TTS may flow these together. Naturally there should be breaks to make it clear there's a table.
… I will provide examples next week.
PaulG: Maybe our demo is not so illustrative of the importance; we want to sell the importance of these features.
Irfan: I think it was wider than pause or break (at least for some big platforms).
… I think it's a good idea to provide a clear use case for each feature.
PaulG: I'm not understanding what's different between the SSML that they already implement for the Web Speech API.
Irfan: I think we should provide some clear examples. We may have more success with other platform vendors.
… Another Explainer perhaps?
Alan: I'd watch saying the word "pause" as that's a Web Speech standard word.
matatk: we have to present a compelling scenario to them, which we've done, but if we can update it fairly quickly and push it to them again maybe it will help
… the opt-out idea is very good. Getting those examples and focusing on the SSML we need with WHATWG will be important
Irfan: I spoke to manish who hadn't received the AT vendor letter