W3C

Protocols and Formats Working Group Teleconference
30 Oct 2014

Agenda

See also: IRC log

Attendees

Present
Janina_Sajka, John_Foliot, Michael_Cooper, Bo_Campbell, Mary_Jo_Mueller, Fred_Esch, Kaz_Ashimura, Joanie_Diggs, Rich_Schwerdtfeger, Jon_Gunderson, Shane_McCarron, Francois_Daoust, Katie_Haritos-Shea, Markku_Hakkinen, Marc_Johlic, Daniel_Davis, Hisayuki, Taku, Hiroto, Shoko, Kazuhiro, Yosuke, Kaz, Daniel, James_Craig, James_Nurthen, Mark_Sadecki
Regrets
Chair
Janina_Sajka
Scribe
jongund, Mary_Jo

Contents


<trackbot> Date: 30 October 2014

<jongund> scribe: jongund

JS: Lunch is from 1-2
... Meeting with digital publishing after lunch
... SVG and HTML this morning
... People please introduce themselves
... We are 4 minutes early

<JF> http://www.w3.org/WAI/PF/media-a11y-reqs/

<JF> http://www.w3.org/WAI/PF/HTML/wiki/Media_Accessibility_Checklist

JS: We need a bigger room next year
... It will continue to get bigger with new initiatives
... Most of our work with the HTML task force

MC: Nobody has called in yet, so we don't need the bridge

JS: JF setup a meeting at Stanford on media accessibility
... We need a requirements document
... Some of the technologies are older than the web
... There are some new version based on the old technology
... Planning on publishing as a W3C note
... We almost have support for almost everything in the document, SF can give more details
... The WebTV working group has their own specs
... I am very proud of the requirements documents and a showcase doc
... We want to see this delivered to people around the world

KAZ: Review WebTV accessibility requirements

SF: I pasted in the URLs to the requirements, including the checklist
... The creation of accessible media, how people with different with different types of disabilities use content
... we looked at the requirements of the system, like the need for controls
... This is a requirement that we need, based on the needs of people with disability
... On monday we identified a need for clean audio, filter out background sounds for example
... Another requirement is time shifting, video description ....

<kaz> tv use cases/requirements for accessibility

SF: The supplemental voice describes something in the video, it traditionally is a human voice, it is a third voice that needs to fit in the quiet of the other speakers
... It could be voiced text, instead of a human
... 120 wpm is typical speech rate, people who use screen readers can often listen to speech at higher rates
... those are some of the requirements we are looking at, we do not have implementation of all these

JS: All the different types of MAUR....
... We need examples that people can test their implementations

KAZ: Thank you very much, the IRC has the second round use cases
... One of the key features of WebTV is very important, we talked to JS and SF about clean audio, on WebVT wiki
... We already generated many use cases, we need to reconsider from accessibility use cases

Yosuke: Today we have second largest TV, we have a guest today from Japanese government and network
... We think it is difficult to explain to production guys, challenge to us to raise awareness
... We need legislation to require the use of these features
... We need something to make this happen
... How do you approach governments, or your out reach plans, like Comcast
... If we can come up with good ways to out reach plans

JS: This group would not be talking to the government, but people we know do that, yesterday we had an FCC commisioner here
... As content migrates to the web, there is no restrictions on "spectrum", does not have traditional limits of over the air broadcasting
... We have a solution if government says you need to do this, clean audio is a first step
... JF described it as a high contrast version of audio
... People who benefit are people who are hard of hearing
... The dialog of the news caster or sports announcer is much louder than the background noises
... We need to write up scenarios , build profiles on your sound board, you can create a clean audio during production

JF: We also looked at other standards, we mapped to a AAA requirement of WCAG 2.0, SC 1.4.7
... ...reads SC 1.4.7 ...

<JF_> http://www.etsi.org/deliver/etsi_ts/101100_101199/101154/01.09.01_60/ts_101154v010901p.pdf

JF: There is a normative ETSI
... It is not so much the primary audio track, if you had multi-channel, this would be a secondary track in parallel, user agents can allow use to choose

JS: BBC does it this way

JF: We now go back to the

Yosuke: Do you have plans on best practices?

JS: Yes I would like to do that

Yosuke: That would really help

<kaz> WCAG 2.0

Katie: I am wondering if there is no restriction on "spectrum", but is there restrictions against bandwidth, if this bandwidth, deliver this
... The size and bandwidth is important

JF: I think you are right, deliver over the web, HTML 5 you can query on information including bandwidth to get content
... We have not specified that, we can go back to the technical people on that

JS: That is more about user agents allow user to choose
... That has not be focus

JC: Content negotiation is already apart of HTML5
... The priority is highest quality based on bandwidth

JF: Users still might want a choice from A or B if B has the clean track

JC: There is already a priority for audio descriptions

JF: That needs to be part of the media query

RS: Are they using a finite ..., BBC breaks it out as a separate track?

JS: Yes

MC: We have someone on the phone

JF: The ITSC document is a mixed at the receiving end, lets not worry about the delivery method, it can be done at either end

RS: Is it a separate voice channel

JF: There is 2 audio track, track 2 has the clean audio

JS: We need to get the BBC involved, since they have done it

RS: I want to understand what you are asking the producers to do

JF: Either a separate media source, when it is available give it to me

JS: One of the things we ask to send to another device, you could send the audio to a filter device
... Something that could be used in movie theaters

Yosuke: JS provided clean audio to WebTV group, we had a joint meeting
... One of these screens it has different code or time stamp, basic way a media controller API...
... Lets assume they all have the same time stamp, how do we mix these, try to provide clean audio...
... The time code can be different causes a disconnect between visual and audio

<Zakim> jcraig, you wanted to ask why the time code is different

JC: Why is it that the time code different?
... You want them at the player ...

Yosuke: These are just streams, like live broadcasts

JF: How does this different from picture in picture in sign language

Yosuke: It is the same, we need additional information

JF: in HTML5 right now, you can have your support resources "in band" or "out of band", so you might have your audio description that would need to be coordinated
... This problem is a little bigger than accessibility, we may not be the group to give the answer

<jasonjgw> michaelc, I'll ask for room 1234 when you go ahead.

Shane: It is completely different for text and audio streams

JF: How is it identified
... If you are identify what these alternative sources are

Shane: There is a label on it

JS: My take away is to create some examples

JC: We should start with non-live content

Yosuke: We are starting to deal with that issue, we would like someone to collaborate with, we think it is essential for clean audio

JF: We can support the use case requirement

<kaz> fyi, there is another tv requirement on synchronization as well

JC: You should raise that in the HTML working group

JS: We may need an issue in the task force

JC: I don't think it is specific to accessibility

JF: But it is important to accessibility

JS: Is mark back?

MS: SVG can meet from 10:00-10:45
... They are where Indie is

JS: We need to get over there

MC: Jason is trying to join, what do we say to him

JS: Go tot he SVG channel
... Thank you Yosuke
... We will keep going on this

RS one on the cue

<jasonjgw> michaelc, they picked up - asked for the name of the guest - I explained what it was and nothing further was said or happened (they calimed they couldn't connect without a name, so I said it was Protocols and Formats working group meeting.)

DDvais: You can describe in the label, but maybe you need a taxonomy

JF: We don't have any right now, but that is a good idea

<jcraig> jasonjgw: PF group of W3C TPAC in Suite 1234

JF: Is there a taxonomy for the aria-label values?

RS: no

JF: A taxonomy for label terms

<jasonjgw> Thanks. What's the SVG conference code?

Katie: That is what I was trying to say before

JC: I am not sure what taxonomy for label

JF: There are 5 values right now

JC: Is that already in the spec

<Rich1> Rich: it is indeed 7841

JF: There is a kind attribute on a track

JC: it should be enumerated

JF: We need to write an extension spec

JS: or errata

Joint meeting with SVG

<jcraig> https://www.webkit.org/blog/3302/aria_and_accessibility_inspector/

<jcraig> https://www.webkit.org/blog-files/aria1.0/africa_large.svg

<MichaelC> Minutes of joint meeting with SVG

<MichaelC> close action-1360

<trackbot> Closed action-1360.

Joint meeting with HTML WG

<MichaelC> Minutes of joint meeting with HTML

Joint meeting with DPub

<MichaelC> Minutes of joint meeting with DPub

Further joint meeting with HTML WG

<MichaelC> Minutes of further joint meeting with HTML WG

Meet with APIP

<MaryJo> scribe: Mary_Jo

<MichaelC> scribeNick: MaryJo

MH: IMS Consortium has QTI/APIP which is an accessibility specification - an XML specification for accessibility.
... QTI has different item types defined, multiple choice (single and multiple selection), matching tasks, question, fill in the blank, etc.
... There are prototypes available for many of these constructs.
... These are mapped into HTML constructs. The QTI model is used for many age groups - K-12, collegiate, post-graduate.
... Demoed some of the prototypes.
... Uses MathML, but also uses some images.
... There are technical challenges in presenting a question that has a drop-down listbox for choosing an answer in the middle of a sentence and how that is best presented to AT.
... When a math equation or expression is spoken, pauses in the speech may be needed and the length of the pause may need to be longer or shorter depending on the use case. CSS may need to be used to adjust the pause, but there isn't such a construct in CSS.
... for speech.
... Some questions contain a passage with multiple blanks and there are different lists of selections for each of the blanks.
... So the challenge is to connect the part of the passage with the possible answers. One possible implementation is using a WAI-ARIA region.
... There will be usability testing done to determine what is the best solution to each of these examples.
... When you hear the paragraph, you hear the blank announced. Then you navigate to the question, select an answer and can hear the sentence again with the answer filled in.
... Will share the examples with the group when they are ready to get feedback.

CS: You might consider partnering with IAAP on this.

MH: Single select multiple choice maps to radio buttons and multi-select multi-choice maps to checkboxes but these have different keyboard operations.
... Instructions will cue how many items to select.

FE: What about the amount of time given? Aren't these timed tests?

MH: There are time accommodations made for those who need it.

Joni: If you want a different keyboard interaction model for the multiple select versus single select you'd have to have a different role and get AT's to implement it.

MH: We're thinking about presenting the multi-select as checkboxes and give them a different role. We are creating prototypes to test the different approaches.
... We are also testing the different presentations of the reading passage and questions with replacement words to see how well that is accepted vs. the inline drop-down list.
... We'll be presenting our findings at CSUN in March, if our topic is accepted.
... For editing a passage, there is also punctuation checking and questions on what is the best punctuation which presents different challenges.
... Current tests have several paragraphs of a reading passage followed by several questions so moving back and forth between the passage and the question is definitely a challenge.

BC: You might be able to find some issues that can help drive back changes into the design of the testing.

MH: We're going to collect data over time to see if changing the layout of the information doesn't change the nature of the question and what they are trying to test.

FE: Do you use charts and graphs in the tests?

MH: Yes, there are. From simple line graphs where it asks what the slope is to more complex graphs and tables that ask more complex questions.
... Adaptive testing may provide tactile versions of these graphs for the student to utilize.
... Longdesc can be used as an attachment to the image as an embossed file.
... A text description can also be used in the Longdesc. There can also be descriptions for the test administrator or teacher.
... We are adopting a diagram content model that has more information than current HTML supports.

JG: How many questions do you do the diagram content model for?

MH: New questions are all being handled in this way to provide alternate formats wherever needed.
... There is a way for SVG to be converted to a format for embossed output.
... Image accessibility - you can have a variety of versions of the same image.

Joanie: Fill-in-the-blank with choices below can cause issues for low vision users. Even with live regions - if they are spoken after selections are made, the braille rendering of the filled-in sentence may create issues.

MH: For live region example, when you make your selection the sentence is updated by filling in the blank.

Joanie: The best solution may be an update to WAI-ARIA and the PF working group can help with that.

MH: There are issues with translating APIP to HTML and there are some very bad implementations out there.
... I want to get this implementation right. Tactical solutions are hacks, so we want things to work correctly.
... There is an issue of text redundancy in APIP. An example is correct pronunciation of words. There are better ways to do it than copying all of the paragraph with pronunciation hacks for screen reader users.

CS: Is there still a problem with old browser versions being used?

MH: We will be providing platform and user agent preferences for these accessible versions.

Joanie: There will have to be a two-pronged approach. There may be things that aren't perfectly implemented now, but we don't want to shut the door on future implementations that are better and reduce the amount of hacks out there.

MH: We aren't yet saying these implementations we have are the right or only way to do things.
... We'll be doing user testing to find out what works best.

<Ryladog_> http://www.w3.org/TR/pronunciation-lexicon/

MH: This work falls under books that are used for assessments. For navigation it would be best to be able to call out the reference blocks, its associated question block and be able to easily navigate between them.

CS: There may be an aside named 'passage' and then the question.

MH: Students want to test with tools that they use on a daily basis - their own screen readers, etc.
... We want to build in compatibility with their existing AT.

JG: WAI-ARIA does have limitations that don't account for the special circumstances of testing.

MH: There are people from testing organizations, Pearson, etc. and want people from AT communities to be a part of this.

JS: Propose a CSUN meeting about this.

MH: There will be a panel at CSUN on this topic.
... Last year APIP was controversial at CSUN, and we want to present what has changed and what we have done at the upcoming CSUN.
... Next version of QTI will be called AQTI - accessible QTI.
... Hashed through a lot of issues with the U.S. Access Board. We want to extend the amount of time for the CSUN panel this year.
... We have large populations of students using this specification.
... Are doing some prototyping using W3C's TTS API, using Benespeak. We have a JavaScript library with highlighting, TTS, CSS3 markup to used to highlight text as it is read aloud.
... There is adaptive testing with people with disabilities also being done.

JS: Would like a F2F on location to see what you've done.

RS: This work was going to be done in a digital publishing sub-group for WAI-ARIA. However the testing is a modular extension of what they're working on - a lot more involved.

MH: Some semantics can be pushed into the ePub model, but some things we don't yet know how to do correctly.

JS: Don't need a new group formed on this yet. We'll need use cases.

Joanie: Would like to pitch in to help when cycles are available.

RS: Could devote some time on the Monday calls, but we need to finish WAI-ARIA 1.1 so we can move on to this new work.
... We'll have to talk about the scope of work.

MH: aria-label is a very popular hack. aria-hidden is also something being used. These hacks will be hard to get undone later. Would like to get QTI markup right first and then the transform can be changed later when the standards are right.

CS: It would be good to be able to skip questions and come back to it later.

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.138 (CVS log)
$Date: 2014-10-31 00:56:15 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.138  of Date: 2013-04-25 13:59:11  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/Yu/You/
Succeeded: s/and/an/
Succeeded: s/#svg/scribe: jongund/
Found embedded ScribeOptions:  -final

*** RESTARTING DUE TO EMBEDDED OPTIONS ***

Found Scribe: jongund
Inferring ScribeNick: jongund
WARNING: No scribe lines found matching previous ScribeNick pattern: <MaryJo> ...
Found Scribe: Mary_Jo
Found ScribeNick: MaryJo
Scribes: jongund, Mary_Jo
ScribeNicks: MaryJo, jongund
Default Present: JasonJGW, Suite1234
Present: Janina_Sajka John_Foliot Michael_Cooper Bo_Campbell Mary_Jo_Mueller Fred_Esch Kaz_Ashimura Joanie_Diggs Rich_Schwerdtfeger Jon_Gunderson Shane_McCarron Francois_Daoust Katie_Haritos-Shea Markku_Hakkinen Marc_Johlic Daniel_Davis Hisayuki Taku Hiroto Shoko Kazuhiro Yosuke Kaz Daniel James_Craig James_Nurthen Mark_Sadecki
Agenda: http://www.w3.org/WAI/PF/meetings/tp2014.html
Found Date: 30 Oct 2014
Guessing minutes URL: http://www.w3.org/2014/10/30-pf-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]