Audio Description Community Group

<scribe> scribe: nigel

This meeting

Nigel: [iterates through agenda]
... Any other business?

group: [no other business]

Issues

Nigel: We have 4 open issues, only one of which is really important

Issues on ADPT

Nigel: Issue 4 is typos, following Cyril's review
... Issue 3 is another editorial task fixing an HTML fragment id
... Issue 2 was raised by me, and its about removing a note about DFXP, which is again mainly editorial
... Issue 1 is the main one I wanted to discuss

https://github.com/w3c/adpt/issues/1

Nigel: My first list of features only included media time, but the first edit of the document includes media time,
... clock time and smpte timecode.

Tracey: Editing AD scripts is easier from a practical point of view if they use smpte timecode

Nigel: My reference was that IMSC for subtitles only supports media time.
... Also, even with media timebase you can still use frame based time expressions.
... The question is then whether the reference point is the beginning of the media (e.g. the equivalent of, say 10:00:00:00
... in timecode), or the zero timecode point.
... There have been issues reported in the past about the use of smpte timecode especially in discontinuous mode,
... if the burned in timecode in video doesn't reflect the timecode data, for example. Also the complexity of the
... AD files themselves would be increased because relative times cannot be used inside the document, so every
... time would have to be replaced by a fixed non-relative smpte timecode label.
... I discussed this with Yella Umbrella a few weeks ago and the conclusion was they could live without SMPTE timecode.
... My proposal is to simplify the spec as much as possible and only support media timebase, and possibly include
... some examples of how they could work with SMPTE timecode-like values in them.
... I could propose use of some metadata to signal the basis of time expressions.

Óran: I tend to agree, it's unnecessary to include smpte timebase in the document.

Nigel: Is there a need for clock times too?

Óran: I can't imagine there is for AD.

Tracey: Not for AD

Nigel: OK we're in agreement here for now, so just for the minutes, we have a (small) consensus to support media timebase only,
... so we can record that and if anyone disagrees let them come back via the issue or the reflector.
... That's all the issues.

Status with TTWG

Nigel: Sorry for missing this before!
... W3C process means that only Working Groups can publish Recommendations, i.e. standards.
... Timed Text Working Group (TTWG) works on TTML specifications, and I also Chair that.
... I proposed that TTWG adopts this and that proposal was accepted, for work in 2019.
... The TTWG Charter is being revised to include it, hopefully to be confirmed in May or June.
... That Charter will refer to our document as the basis.
... So we have a few weeks or so to get the document in as good a state as possible.

Speed scan through document

Nigel: Any initial high level comments on the document?

Óran: Having trouble navigating through the page.

Tracey: I will go through it with Óran.

Nigel: Is there anything we can do to help that?

Óran: The way the document is labelled and the navigation system works is a bit cumbersome.

Nigel: I'd like to know more about that.

Óran: Like a lot of web pages, their layout is not very well formatted for use with screen readers so navigating headings

scribe: Óran: can be a bit tricky. I encounter this quite a lot.

Nigel: It would be super-helpful to be able to feed any issues back to W3C and to the team maintaining the Respec tool.
... [iterates through document]
... Seems like the Introduction should go as close to the top as possible.

Óran: Yes!

Nigel: I would also like to move one or more of the examples into the introduction to help readers understand the
... kind of document that is being specified.

Óran: Yes.

Nigel: Then there's a section on Workflow and another on Requirements. Some of the workflow can probably move to
... the Introduction, if it helps understand the scope.

Tracey: Yes, the diagram especially.

Nigel: Then the detail, and the list of requirements I would put into an appendix.

Tracey: Yes

Óran: Yes, I can't see the diagram of course!

Nigel: True - that's why there's a table listing the workflow steps, though it's not a 100% perfect replacement, more an
... extra layer of detail.
... Then we have the Profile section itself, which is very short, and just references the Constraints section.
... The Constraints section is the main body of the specification. It describes the encoding, namespaces, extensibility,
... synchonisation, profile signalling, and most importantly of all, the set of permitted or prohibited TTML features.
... Those features are the most important thing! I'd hope to move those closer to the top of the document too.
... In IMSC, for example, they're in section 6, here they've got down to section 9.
... then there are appendices, the examples, Web Audio mixing, acknowledgements and references.
... Any other observations about the document structure?

Tracey: It's the most detail I've seen.

Óran: Is embedding audio description something to be discussed?

Nigel: I think the scope of this is upstream of embedding - the goal here is to define the script and the audio mixing,
... which can be used to generate the audio for embedding. There are already other standards for embedding, so in that
... sense it is not needed.

Óran: Yes

Nigel: I wouldn't rule out in-band carriage of the TTML AD script as a piece of work in the future, to provide the script
... and audio to a player for native playback or screen reader use, for example.
...
... Okay, that's our agenda, and we're a bit over time. I will turn those into editing actions and follow up with John Birch.
... If he can't do the editing then I will try to.
... Thanks for joining. [adjourns meeting]