Spoken Pronunciation Task Force Teleconference

03 November 2021


Alan, Dee, IrfanA, janina, JF, mhakkinen, NeilS, PaulG, Roy, SteveNoble

Meeting minutes

Agenda Review & Announcements

PaulG: standard agenda along with GitHub issues. Mostly Debriefing form TPAC

TPAC debrief

PaulG: great experience. especially working with top 5. browser vendors are pretty open to implement without having any preferences.

PaulG: SSML as a first class object is preference.

PaulG: let's pick JSON single attribute model.

PaulG: if anyone has any opinion on this, please share

janina: ack j

janina: have no problem with Paul's approach. we have smother way to get there. However, if browsers do not care then we need to think whats the best way for authoring perspective. then the tool can transform that into json.

PaulG: we still have some concerns about the performance and AAPI mapping. once we do, it is still going to take time. in authoring we could have tool such as read a loud or some extension.

janina: isn't the AX tree a real issue regardless of any approach.

PaulG: once data gets move to ax tree, its less concern to the technology that is consuming it. only concern is how does that relate to live region or browser updates this info. having the extra layer of pronunciation information there, that not user wants

PaulG: once you get it to ax tree, fewer people would care

janina: as far getting stuff whatever ax tree solution we do, there is no difference whether the source is single attribute or multi attribute

janina: getting it to ax tree there is no performance skip. this is what heard

NeilS: my understanding is the same that it doesn't matter how you author this. I want to bring another alternative that was discuss such as CSS like approach which is particularly useful.

NeilS: you can always do inline-css if you have bunch of different things. this came someone from TAG and can come up again. it does seems to have an appeal.

NeilS: I do not remember if it was ever discussed and if it was, why did we drop.

mhakkinen: i agree with Neil. Css was not something came up positive in our gap analysis

NeilS: this could be something css like

mhakkinen: one mistake in my mind with the single attr approach was to have an reference of style block or speech block and using it frequently. css was really extra coding. could we come up with some approach which is css like and can be parsed easily.

NeilS: I am not trying to advocate CSS approach but it does have an appeal.

PaulG: there was a conversation of dictionary approach which was short sighted. it doesn't solve anything

PaulG: it was interesting to bring up but it doens't resolve anything other than phoneme

PaulG: single attr approach does have couple of extra benefits.

PaulG: because if everything is one logic, i think as an implementation decision, it seems cleaner.

NeilS: I think thats a good point.

PaulG: json as an external resource is easy to parse

PaulG: which could make our presentation code more useful

janina: playing devils advocate, though I support it.

janina: where are we in 10-15 years

janina: how hard is the conversation part. will we still be able to work

mhakkinen: I'd think that it can be walk back with all the necessary information available

PaulG: there will be some authoring decisions

PaulG: unless you use a tool which has only one way to author, this would be like a general appraoch

janina: what I am hearing that there is path forward which may have more human mediation that you may want but there is a path forward for sure

PaulG: some people spoke against single attr, they may have some say on this. even if some one is using braille or screen reader to author may have different opinion.

PaulG: ax tree may not need all the information in multi attr case with nested pair. it may not be necessary

mhakkinen: I was going to suggest that we went through some cycle at ETS and updated UI for authoring the ssml pronunciation. perhaps we can have a demo sometimes to show how did we handle this.

PaulG: may be we will sit on it little bit. Irfan and I will be having a discussion, what should we take it to our presentation in access higher ground conference

janina: you can certainly quote it as w3c process

GitHub Issues

janina: we could start walking through them and action since we have a little path forward.

janina: we can close some github, update our document and publish it

PaulG: next week we come back and see if we still have the same direction to move forward

PaulG: we can then decide to close some actions

janina: there is a APA sponsored symposium funded by EU 'Shake the future', next Wednesday. we may need to skip next week and following week.

PaulG: which means we wont be able to have this meeting until 1st.

janina: we take next 3 weeks off because of various reasons so we would returned in Dec 1st and take last two weeks off in dec and returned in Jan 2022

mhakkinen: where is the info for symposium for next wednesday.

janina: I do not have it right now but we can have it in APA call or need to ask Josh about it

<Roy> https://www.w3.org/WAI/about/projects/wai-coop/symposium1/

<janina> https://www.w3.org/WAI/about/projects/wai-coop/symposium1/


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


Succeeded: s/shake/shape