– DRAFT –
Accessible Platform Architectures Working Group Teleconference

<Judy> [jb: regrets for the beginning of the call, will join as soon as gets out of another call]

<Joshue108> Off-page link for the captionmax transcript: https://www.streamtext.net/player?event=WorldWideConsortium

Joint working group meetings and break-out sessions planned for TPAC 2021.

<janina> https://www.w3.org/WAI/APA/wiki/Meetings/TPAC_2021

janina: some TPAC meetings have moved - so be sure to check the schedule

scott_h: +1 to Janina's work on scheduling

janina: connection to RQTF...besides timed text, maybe personalization is another area

janina: metadata on media accessibility options (captions, video descriptions, etc.) - these options often vary from one streaming service to another

<Raja> So, what can be standardized?

jasonjgw: EPUB spec does include metadata requirement for the item, plus push through to EPUB distributors

<Joshue108> SN: Raja asked a question in IRC on standardisation - what can be standardised?

<Joshue108> JS: Raja can you please restate question?

<Joshue108> RK: Yes, I was talking about meta data but I’m not an expert

janina: One issue is simply figuring out who to include in the conversation'

Synchronization Accessibility User Requirements.

jasonjgw: A lot of work has been going on

jasonjgw: Believe we have addressed enough of the outstanding issues

jasonjgw: Have incorporated the new BBC white paper and an editors note around the issue of accuracy in ASR

jasonjgw: Should we request publication at this point to get wide review?

Joshue108: Yes...it seems good to go. Just would like Steve to review

Raja: Regarding ASR change it is improved is looking beter

Raja: Need to be careful in how we decide to leave this since we do expect more changes

<Joshue108> SN: We have the doc in good shape

<Joshue108> We do have some questions - addressed in editors note etc

<Joshue108> The issue around ASR is between what is the real world experience is like

<Joshue108> It is seen to be lacking in standard Zoom calls for example - but this is the MVP

<Joshue108> You can pay more for premium product

<Joshue108> There is variability

<Joshue108> Most metrics are from lab type environments

<Joshue108> Different from e’day life

<janina> https://www.w3.org/Consortium/Process/

<Raja> Yes, the notion of variability and premium products won't change soon, and is important to share

janina: The type of document we are working on is a "Note"

janina: This is a "working draft" meaning that it is not done, and since it is the "first" working draft we will assume that there will be another draft

janina: We work on it until we are satisfied

janina: Suggest that we reference the recent FCC posting

janina: Otherwise it will look like we have ignored it

<Raja> +1

jasonjgw: Around a recent FCC announcement on the ability of a vendor to use ASR

<Raja> CaptionCalls got certification to supplement human captions with ASR yesterday

jasonjgw: We could simply cite the reference

<Raja> Other caption providers got permission several months ago -- Olelo and another provider I can't remember at the moment

<janina> https://www.fcc.gov/document/ip-cts-asr-only-authorization-granted-captioncall

<Raja> Just having permission doesn't carry force of law

Raja: Yes...the new order allows ASR if there are not enough human captioners

Raja: Other providers also received such permissions earlier this year

Raja: This may help support the case that ASR is seen as becoming more acceptable

<Raja> https://www.fcc.gov/document/machinegenius-certified-automatic-speech-recognition-ip-cts

<Raja> https://www.machinegenius.com/

Judy: The understanding is that this applies to limited circumstances, so we want to look at this more seriously

<Raja> Clarity and ASR: https://docs.fcc.gov/public/attachments/DOC-364731A1.pdf

Judy: Also W3C is international and the FCC is just one country, plus ASR is much better normed for English; therefore we want our document more international in its approach

janina: The metrics are indeed controlled, and real life is not a test lab

janina: We are trying to say something about a US agency has approved use in constrained circumstances

Judy: Other parts of the world do not have the same availability

<Raja> ASR will almost certainly be worse for languages used by small populations

janina: Can we continue discussion in email?

<Raja> I agree with the level of caution.

<Joshue108> SN: This note of caution is fine

<Joshue108> ASR sync - is much more quick in generating captions..

<Joshue108> There are issues in quality and accuracy - that’s the main thing.

<Joshue108> JW: This document is not talking a position on ASR just noting it.

<Joshue108> SN: +1 to having a position of caution is good - and we are acknowledging limitations

<Joshue108> JW: We could leave the text untouched and cite these in the next round?

<Joshue108> JS: I’d like to avoid that

Judy: Adding the clarifications will be very valuable -- also refering to language diversity in the country that just issued this decision

jasonjgw: Cited a 2019 paper that did referenced speaker variability

jasonjgw: Are there other items in the note we need to address?

<janina> DRAFT RESOLUTION: RQTF requests APA conduct a Call for Consensus on publication of a First Public Working Draft of the Synchronization Accessibility User Requirements (RQTF) https://raw.githack.com/w3c/apa/main/saur/ once our Editor's Note is updated by agreement to address recently published guidance from the U.S. FCC

<Joshue108> SN: <reads draft resolution>

Resolution: RQTF requests APA conduct a Call for Consensus on publication of a First Public Working Draft of the Synchronization Accessibility User Requirements (SAUR) https://raw.githack.com/w3c/apa/main/saur/ once our Editor's Note is updated by agreement to address recently published guidance from the U.S. FCC

<Joshue108> JS: Many thanks to Steve

Judy: thanks to captioners and interpreters on the call

janina: Josh can mint the URI