W3C

– DRAFT –
Spoken Pronunciation Task Force Teleconference

18 May 2022

Attendees

Present
Alan_, Dee, Fazio__, iali, janina, PaulG, Sarah, SteveNoble
Regrets
-
Chair
Paul G
Scribe
iali, Irfan Ali

Meeting minutes

Agenda Review & Announcements

<Fazio__> Down syndrome affects kids' ability to learn in different ways, and most have mild to moderate intellectual disability. https://kidshealth.org/en/parents/down-syndrome.html

janina: we are expecting to meet DPub and Internalization at TPAC.

<Fazio__> Down Syndrome - much higher receptive skills than expressive skills

PaulG: We will need to discuss our proposal with ARIA.

janina: we may need to have another conversation with them prior to TPAC

<Fazio__> Augmentative Alternative Communication Devices AAC is critical early intervention - Down Syndrome

mhakkinen: got invited at the CCSO, department of education. meets in Atlanta in the end of the June. Apple and Google and some others including Pearson will be there and one of the topic is pronunciation

janina: you may be interested in one of the item that Paul is working on.

Guest: David Fazio (from COGA)

Fazio__: introduction*

Fazio__: link to an article with a relevant quote

https://kidshealth.org/en/parents/down-syndrome.html

Fazio__: we did some study on down syndrome.

Fazio__: the part of expressive difficulty is some king of imperative speak. because of the multiple modality of the devices

PaulG: using aac or interactive with AT, where the correct pronunciation is a challenge to learning process, can you talk something about that?

we get two side of this, there are lots of context where ATs are wrong. curious if there is anything about receive info from spoken or TTS engine that would further complicate the learning process.

Fazio__: they will create the picture of action or activity. person with disability articulate to the audience the expressions. lot of times parents create the specific images to respond them. There is not much research about it. But it needs to be clear in order to make things easier

janina: when the custom acc created, we made easy for people to author for better pronunciation which is useful.

<mhakkinen> The Voice Keeper enables creation of TTS from your own voice, AAC is one applicaiton: https://www.thevoicekeeper.com/

Fazio__: I have an entire binder with the instructions. I can provide this to the group.

mhakkinen: there was a research for the smoke alarm that would have a voice of the parents. personalization of the voice can be very helpful.

PaulG: thanks for joining David. please share the research info that you have.

Action Items

PaulG: day time issue is still open

<PaulG> https://github.com/w3c/pronunciation/issues/101

<PaulG> https://github.com/w3c/pronunciation/issues/110

PaulG: do we think we have all the use cases from Children group?

https://mlc.learningstewards.org/help/#Settings

janina: just so you know that we are writing a use case which will go further in two weeks. WayAdopt is the group. It used to be called personalization which was part of COGA. We should be coming out of the CR before TPAC. there will be two CR. first CR with data-

mhakkinen: sounds like halfway (pathway) to us.

janina: David has given us some interesting use cases.

PaulG: we will need to put these use cases in our documentations. Good to have all the use cases asap.

PaulG: need to update the use case document.

PaulG: need volunteers to help in this document. as an editor or author

Github Issues

https://github.com/w3c/pronunciation/issues/111

<janina> https://lists.w3.org/Archives/Public/w3c-wai-ig/2022AprJun/0037.html

PaulG: govt agencies have acronyms, they are gong to be spoken to you at some point and if you have learning disability it is important for tts to have the mechanism to speak these acronyms in the same way.

PaulG: we can respond to this ticket.

Dee: need to comment on authored vs spoken pronunciation in the context of the assessments.

the intended pronunciation is important in assessments so they are hearing the correct thing.

PaulG: some examples like triple one or double seven, these are very challenging use cases for tts

Dee: some words are pronounce in different way then they are authored in the assessments. If the SR reading them in one way and tts in another way, there could be a gap

PaulG: priority is what the author dictate.

mhakkinen: its like ccs !important property where user wants to override everything except the stuff that is marked as important.

mhakkinen: we talked about the implementation guide for the AT vendors

janina: its a good idea

PaulG: we can have a github with the AT label

PaulG: you get arrow in HTML and coding it in unicode char. easier approach would be a literature font where you can use some glyph when the AT reads, it will read the text and not a graphic. need to test it

janina: not sure how it works in the braille

PaulG: I want to get a literature font example and will send it to Janina to test it in Braille

Alan_: I was thinking of this issue which makes it clear as far as symbol is actually means.

PaulG: it would be a substitute instance. without the specs if it works in the braille, its good

PaulG: not necessary most author friendly mechanism though

Other Business

Minutes manually created (not a transcript), formatted by scribe.perl version 185 (Thu Dec 2 18:51:55 2021 UTC).

Diagnostics

Succeeded: s/spoke/smoke

Succeeded: s/dat/day

No scribenick or scribe found. Guessed: iali

Maybe present: mhakkinen