W3C

Timed Text Working Group Teleconference

25 September 2025

Attendees

Present
Andreas, Atsushi, Chris_Needham, Cyril, Gary, Nigel, Pierre
Regrets
-
Chair
Gary, Nigel
Scribe
cpn, nigel

Meeting minutes

This meeting

Nigel: [Recaps the agenda]
… Anything else to cover, or points on the agenda items?

(nothing)

Topic IMSC

Make the reference to The Unicode Standard undated w3c/imsc#618

github: w3c/imsc#618

Pierre: The ISO platform doesn't offer URLs to the latest edition of the standard

Gary: Unless we want to refer to Unicode instead?

Nigel: Why not refer to Unicode?

Pierre: Other specs refer to it
… ARIB refers to collections, and I want to make sure that's in the Unicode standard, not only in ISO

Nigel: I checked the ISO website, they have a mechanism for adding dated and undated links to ISO specs, but I couldn't find access to the database they use

Pierre: Could change the reference from ISO 10646 to Unicode and be done
… It's free to download from ISO

Nigel: To answer this, there should be precedent from elsewhere in W3C

Pierre: What about TTML2?
… That references the Unicode spec
… The ISO reference came up in response to ARIB, but everywhere else we reference Unicode

Gary: CSS Selectors links to Unicode. The two are functionally equivalent

Pierre: I'll just reference Unicode

Gary: Sounds good

Nigel: Sounds good

SUMMARY: Change 10646 reference to Unicode

Improve the ja character set per ARIB feedback w3c/imsc#614

github: w3c/imsc#614

Pierre: Atsushi, if I accept both your comments, are you happy with the PR?

Atsushi: The comment about ARIB was just a suggestion

Atsushi's comment

Atsushi: The comment relates to the suggested change above.

Pierre: I accepted the suggestion. Is the PR ok now?

Atsushi: Yes

SUMMARY: Atsushi's comments accepted

Wide Review

Pierre: I think we're done with wide review. There's a potential editorial note by APA

<atsushi> wide review tracker

Nigel: I don't think we've had all the horizontal review feedback yet. Not clear if i18n is done yet. We have review feedback from ARIB

Pierre: i18n has completed the review

Nigel: It looked like APA is considering feedback

APA tracking issue for IMSC 1.3

Nigel: Issue 116 is closed. They want us to think about assistive technology users. Good to address as soon as we can.

Pierre: Their concern is how to make an HTML rendering accessible

Nigel: They are concerned about how it relates to CSS vertical-align. You could have a situation where whatever is rendered in a screenreader the styling is lost

Pierre: The fallback is to not have the superscript, which is fine

Nigel: The sub and sup elements could be useful here. Is it OK for a rendering implementation that maps IMSC to HTML to use those instead of font variation position?

Pierre: I'm leery of including very specific HTML rendering recommendations, as it changes often.
… The browser bug threads say use vertical-align, for maximum compatibility

Nigel: We don't want to get caught into that debate, but adding a note that vertical-align won't report property might be all that's needed

Pierre: I'm OK doing that if it resolves their concern

Pierre: They should propose the Note

Nigel: Can ask APA about adding such a note

SUMMARY: @nigelmegitt to ask APA if adding a note about implementations using vertical-align would address their concerns

Response to ARIB

Pierre: There were two comments not against IMSC1.3 specifically. Let's figure out how to respond

Nigel: Yes, let's you and I do that

DAPT

Nigel: HR feedback: Privacy said no concerns, i18n opened some issues, and no feedback from other groups

Use zxx in place of an empty daptm:langSrc w3c/dapt#322

w3c/dapt#322

github: w3c/dapt#322

Nigel: Feedback in w3c/dapt#321 is that there's a tag for non-linguistic content
… The zxx language tag means what we want here
… I opened a PR, to day well-formed rather than valid. And previously the default was empty, now it's zxx, no semantic change.
… Fuqiao said we could use und instead of zxx
… Do we think und would make more sense? Then we'd need a rule that no content has a langSrc of und, and introduce an error state
… Should we change the semantic, and require users to define it?

Cyril: We want to avoid creating wrong documents
… The attribute is inheritable. I wonder, if the root isn't specified it'll have a default value, overridden at paragraph level
… Don't know if I have a preference for undefined
… It's easier to detect if the document was mistakenly left to the default, when it's und, rather than zxx

Nigel: I think that makes sense
… zxx only makes sense if it's audio description, but probably wrong for other types of transcript. Don't want to introduce a normative requirement on registry entries
… Proposal is to change the default to und, and add a constraint that all completed text objects have a value that isn't und. Any other views?

<atsushi> +1

Pierre: So you'd require the langSrc attribute to be present?

Nigel: Yes, and the computed value for text shall not be und.

Pierre: Forcing people to put something where they might not understand, they'll put in bad values when they don't know

Cyril: For other attributes, I'd say yes, but you'd know the language you're authoring against

Pierre: If the tool forces you to put something, and you don't know about the weird corner cases, someone might put in a value just to stop the tool complaining
… Is there a way that if you say nothing, it means "i don't know", no value if the attribute isn't present

Cyril: Either put the default as undefined, and have semantics to say it should be set, or xx

Nigel: Or include in the registry for represents, have a known value of langSrc to use in the original transcript

Pierre: I'd recommend that the absence of the attribute means und

Cyril: Yes

Cyril: So we're saying the default is und but doesn't need to be overrridden by an actual language

Andreas: It would be more transparent to have it specific, and an explicit value of undefined
… Agree on having the default be undefined

SUMMARY: @nigelmegitt to update the PR to make the default "und"

WebVTT open issues

Gary: Can skip this given the time available, TPAC is more urgent

TPAC 2025 planning

Nigel: I've made a wiki page

TPAC 2025 TTWG Wiki page

Nigel: It has a section for participants to describe their arrival and departure. Useful to know who's coming and who's remote
… You should all be able to edit
… There's a section on Goals, moving our specs forward. We have joint meetings, with MEIG and with APA. Future work
… Does anyone have topics to add to the agenda?
… Joint meeting with MEIG, TTWG main meeting on Tuesday. I need to know timing constraints for particular topics, especially for remote participants
… Thursday has TTWG/MEIG/APA joint meeting

Andreas: In TPAC Seville we talked about the DVB intent to specify subtitle purpose metadata. That's now published, in TV Anytime. Could discuss again, see if it meets our expectations

Nigel: Is it TTWG or MEIG thing?

Andreas: one part is MEIG, but a specific part on the TV Anytime metadata model, was discussed in TTWG 2 years ago, this is more for subtitling experts

Nigel: Having some time to talk about advances in implementations or user communities, who's using our specs and what for, could be valuable

Andreas: Yes, sounds good
… Activities around DVB TTML, for example

Nigel: Other topics?

Gary: A bunch of WebVTT topics. I'll be remote, so in the morning Japan time?
… Question is scheduling that vs IMSC and TTML discussion
… Timing-wise US west coast is a bit easier. For me, 9am in Japan would be 8pm locally

Cyril: Not sure if I'll be there in person

Nigel: We don't have many meetings between now and TPAC. Please contribute to the wiki offline
… Gary, you could send email to the reflector to ask on scheduling

Nigel: Should we cancel the 6 November TTWG meeting?

Gary: Yes. I can't make October 23

Nigel: So your last meeting before TPAC will be October 9

Gary: I can do it async

Meeting close

Nigel: We're slightly over time, let's adjourn.
… Next call is on October 9

October 9 agenda issue

Nigel: Thank you everyone! [adjourns meeting]

Minutes manually created (not a transcript), formatted by scribe.perl version 244 (Thu Feb 27 01:23:09 2025 UTC).