15:00:29 RRSAgent has joined #tt 15:00:29 logging to https://www.w3.org/2018/12/13-tt-irc 15:00:31 RRSAgent, make logs public 15:00:31 Zakim has joined #tt 15:00:33 Meeting: Timed Text Working Group Teleconference 15:00:33 Date: 13 December 2018 15:00:37 Log: https://www.w3.org/2018/12/13-tt-irc 15:00:39 scribe: nigel 15:00:43 Present: Nigel 15:00:50 Regrets: Thierry, Gary 15:00:52 Chair: Nigel 15:01:25 Present+ Glenn 15:02:50 Present+ Cyril, Pierre 15:03:45 Topic: This meeting 15:05:00 Nigel: Today we have regrets from Thierry and Gary. 15:05:31 .. There's been action on the TTML profile registry and the requirements so they 15:05:49 .. will be our main agenda topics for today. 15:06:07 .. I don't know if anyone wants to discuss the CSS issue I sent to the reflector 15:06:17 .. earlier about the combination of ruby and text emphasis and where the 15:06:22 .. resulting marks should go? 15:06:46 cyril has joined #tt 15:06:57 Glenn: I don't think that's been raised before and I don't have an immediate 15:07:01 .. response to it either. 15:07:16 .. It is certainly something that the spec should have language to address to I 15:07:22 .. think we'll have some follow-on actions. 15:07:33 .. Do you want to post an issue on TTML2 or shall I do that? 15:07:47 Nigel: I'd probably be happier if you or possibly Cyril were to raise it as being 15:08:05 .. stronger proponents (and more importantly more knowledgeable) about ruby etc. 15:08:16 Glenn: Okay I'll file an issue and refer to your email in archives. 15:08:18 Nigel: Thank you. 15:08:38 .. That means we'll defer discussion of that issue until a later date if we need a 15:08:47 .. discussion on a call about it at all of course. 15:09:07 .. Back to the agenda, I don't think there's anything on the f2f meeting or anything 15:09:21 .. else on CSS. We don't have Gary today so I don't think we'll be covering 15:09:32 .. WebVTT. Is there any other business to discuss or particular points to make 15:09:35 .. sure we cover today? 15:09:50 group: [silence] 15:10:02 Nigel: I should say I have to catch a plane later so I'd appreciate closing the 15:10:14 .. meeting by the end of the hour so I have a bit of extra leeway. 15:10:39 Topic: TTML Profile Registry 15:10:50 Nigel: Thank you Glenn for merging those pull requests. 15:11:06 Glenn: I need to update the PR on alphabetical ordering and make some progress 15:11:16 .. on the other issues. My time has been completely occupied and probably will 15:11:29 .. be until mid-next-week on other matters, after that I expect to get to those. 15:11:35 Nigel: Ok, thanks for letting us know. 15:12:38 .. I see from https://github.com/w3c/tt-profile-registry/pulse that the two 15:12:48 .. issues and pull requests we discussed last week were closed/merged. 15:12:54 .. Is there anything else on this right now? 15:12:57 Glenn: I don't have anything. 15:13:31 Topic: TTWG Future requirements 15:13:49 Nigel: Thank you Cyril for raising the responsive timed text and karaoke issues. 15:14:46 q+ to mention audio subtitles 15:14:57 Cyril: I did not follow the template but I hope that's fine. 15:15:08 Glenn: The templates are for writing follow-on more detailed descriptions. 15:15:11 .. The issue content is free-form. 15:15:32 ack nigel 15:15:32 nigel, you wanted to mention audio subtitles 15:15:49 Nigel: One requirement that has been brought to my attention today at the 15:16:10 .. ITU/EC Accessible Europe event is to support "audio subtitles". I've asked 15:16:21 .. a proponent to raise the requirement as an issue or offered to do it on her 15:16:29 .. behalf (she's not especially technical). 15:16:40 .. So hopefully that will be raised by the deadline next week. 15:16:58 .. For those unaware, this is the ability to trigger text to speech for translation 15:17:09 .. subtitles mixed in with hard of hearing subtitles. It could well be that no 15:17:19 .. spec changes are needed, but I've asked for the requirements to be submitted 15:17:26 .. so we can satisfy ourselves about that. 15:18:51 .. For those unaware, this is the facility to trigger text to speech for translations 15:19:11 .. to support viewers unable to read the translation text, i.e. it bridges across 15:19:27 .. from a "subtitling culture" to a "dubbing culture" for those to whom that provides a benefit. 15:19:54 .. Does anyone want to discuss any of the specific issues that have been raised 15:19:55 .. so far? 15:20:02 Cyril: One small aspect to discuss is condition. 15:20:14 .. In the responsive design use case I tried to use the condition element, and I 15:20:26 .. mentioned in the issue I am not clear how condition behaves in terms of 15:20:48 .. processing and XML processing in general. You want to re-use the same 15:21:04 .. style id and define it differently depending on the conditions. 15:21:14 Glenn: That's illegal in XML. You can't do that. 15:21:18 Nigel: I've raised this before too. 15:21:28 Glenn: There are different ways to do it. You can't really switch regions, but 15:21:39 .. you can switch animations of regions, e.g. moving a region via conditional 15:21:52 .. animation that achieves the same effect. You can also have styles attached to 15:22:04 .. regions which are conditional. You would reference multiple styles, say you 15:22:08 .. want to switch between A and B. 15:22:16 Cyril: You reference both and only one is activated. 15:22:25 Glenn: Exactly. It may not be the most elegant design but technically it does 15:22:43 .. work and I did test this. I will try to comment to the same effect on the issue. 15:22:47 Cyril: Thank you. 15:22:58 Nigel: There's a thread about this probably on a TTML2 issue that should be 15:23:09 .. closed from the review stage back in the summer where I raised the same 15:23:17 .. question and we came to the same conclusion. 15:23:39 .. One thing you can do to simplify it from the authoring perspective is to 15:23:49 .. reference one style from the content and have that style itself reference the 15:23:55 .. conditionalised styles. 15:24:59 .. For example paragraphStyle -> (bigParagraphStyle, smallParagraphStyle) 15:25:11 .. where each of those is conditionalised so the style tree resolution handles 15:25:23 .. the complexity and the content just needs to reference paragraphStyle. 15:25:26 Cyril: Thank you. 15:25:55 Pierre: So I think the question is if the condition mechanism is adequate to 15:26:06 .. satisfy the use cases we have in front of us. We don't have to decide that on 15:26:13 .. this call, but I would ask that question. 15:26:24 .. Since we have had two people raise similar feedback. 15:26:35 Glenn: Just as a note on that point there is no real way that we can have a 15:26:47 .. conditionalised syntax mechanism that makes something syntactically present 15:26:57 .. or not because XML itself would require that feature or you would need a 15:27:08 .. macro processor or something that you run the document through before 15:27:22 .. feeding it to your XML system. I don't view that as very practical. We maybe 15:27:36 .. just need to do a better job with examples in the spec. There is some language 15:27:42 .. that emphasises it is the semantics. 15:27:51 Cyril: An example would help, I searched for one but could not find one. 15:27:54 Glenn: Good point. 15:28:03 Pierre: Maybe the right way to start is to compose practical examples and see 15:28:10 .. where that leads us in refining the spec. 15:28:19 Glenn: Sure. That may be the ultimate outcome of this issue for example. 15:28:29 Pierre: Exactly, especially as we have two participants who support this use 15:28:37 .. case it would be good to put together concrete examples. 15:28:48 Cyril: I agree with what's been said - most of the requirement does not need 15:28:55 .. new things in TTML, just clarifications. 15:29:19 .. One thing I added was the ability to conditionally split content. Is there anything 15:29:24 .. that already can do this? 15:29:37 Nigel: I've thought about this too and have similar use cases. The one suggestion 15:29:48 .. that I think Glenn made some time ago is that there could be conditional 15:30:01 .. br elements, but I don't think that completely achieves the result. Instead 15:30:13 .. you're conditionally changing the timing of presentation, and I don't think I've 15:30:25 .. seen that expressed so clearly before. Though those with long memories 15:30:48 .. may remember a demo I gave at TPAC (maybe 2015?) where the user had a 15:31:04 .. slider to adjust the text size, and depending on the size, the amount of visible 15:31:19 yes 2015 15:31:22 .. text varied so it always occupied one line, but the rate of change was adjusted 15:31:32 .. to keep up with the audio. 15:31:38 .. That was in Sapporo I believe. 15:32:02 .. That could be considered an example of responsive display and I don't know 15:32:08 .. how you could do that using condition alone. 15:32:29 .. Assuming that duplicating a lot of content is undesirable of course. 15:33:03 Cyril: I would be interested to see that demo if you have it. 15:33:13 Nigel: I have it but only on my computer so maybe its one for the f2f! 15:33:27 Cyril: In this demo and the conditional breaking of events, I see some commonality 15:33:39 .. with the karaoke use case in the sense that if you provide more granularity 15:33:53 .. in the timing then the processor can do something more with it. I call that 15:34:05 .. internal timing with markers that are not used all the time. 15:34:17 Glenn: What do you mean by "event"? We don't have that in TTML. 15:34:26 Cyril: For example a p with a begin and end is an event. 15:34:33 Glenn: How about a span with a begin and end? 15:34:43 Cyril: Yes, something that is displayed on the screen for some time. 15:34:51 Glenn: I guess we'd call it active interval or something. 15:35:00 Cyril: You see why I called it "event" because that's simpler! 15:35:02 Glenn: Yes I see 15:35:10 s/active interval/active temporal interval 15:35:29 Glenn: The "event" as you use it is a little different because it implies a semantic 15:35:41 .. consistency or relatedness whereas an active interval is like the smallest 15:35:52 .. timeslices that divide up all the begin and end times that apply in the whole 15:36:06 .. document so you could have active intervals generated by unrelated content 15:36:17 .. simply because they have intersecting activation or deactivation times. 15:36:19 Nigel: Yes 15:36:56 Nigel: That idea of the marker is something I've thought about in the past but 15:37:07 .. never taken any further. The other useful construct to observe or be aware of 15:37:26 .. is the idea of a pseudo selector that affects things in the future vs things in 15:37:32 .. the past, which WebVTT has. 15:37:49 .. That's quite a powerful idea in this context. I'm not clear if it does everything 15:38:00 .. that's needed but it's pretty easy to see how it is useful for karaoke for example. 15:38:38 .. It's a different way to think about the problem, I'm not sure if it addresses the use case. 15:39:00 Cyril: I agree Nigel. The marker is similar to one of the tags in WebVTT where you 15:39:20 .. can set a tag to affect anything, e.g. a speaker, and it does not affect the 15:39:41 .. display. Is it the v tag? 15:39:59 Nigel: I'm not sure if its the same thing but I recall a way to affect the display 15:40:01 .. within a cue. 15:40:13 Glenn: We can add metadata of course in TTML but there are no presentation 15:40:15 .. semantics in that. 15:40:32 Nigel: That's true, I think this use case does want to allow for some presentational semantics. 15:41:08 Nigel: There's another question, which is are we expecting all players to 15:41:20 .. honour this or is it okay for specific players aware of this functionality to do it? 15:42:07 .. Another question about marker is how it fits within the nested span structure 15:42:16 .. that's used in TTML to define ruby presentation? 15:42:39 .. There are loads of questions like that I suspect - I'm not raising it as a blocker, 15:42:45 .. just throwing it on the table as something to be aware of. 15:43:24 Nigel: To add more about the BBC's use cases, we definitely have this vertical vs 15:43:31 .. horizontal kind of issue, also square video. 15:43:44 .. For example think about a video posted to twitter, which was originally 16x9 15:43:51 .. but appears in the preview as square! 15:44:36 Cyril: At least for Netflix there are two aspects that this Responsive Design 15:44:45 .. technology could help. One of them is regarding the use of the Timed Text 15:44:55 .. content in the workflow, having a single document for multiple use cases 15:45:08 .. simplifies the workflow - one processing. The other is the final delivery to 15:45:16 .. the client, where receiving one document makes it easier to adjust. 15:45:31 .. These are two different sub-use cases. You can imagine receiving one document 15:45:38 .. from the studio and delivering multiple documents to the client. 15:45:51 .. The solutions may be simpler than presenting one document in different ways. 15:46:09 .. I just wanted to highlight that. 15:46:12 Nigel: +1 15:46:23 Cyril: In the lifecycle of the document there are two places where this could be useful. 15:46:39 Glenn: This has an echo of an earlier conversation that the group had a very 15:46:49 .. long time ago pertaining to language where we discussed if multiple language 15:46:58 .. versions should be in the same document or in separate documents. 15:47:10 .. I recall at the time that for distribution to rendering clients it is preferable to 15:47:21 .. use a single language inside a document. But we also recognised that at the 15:47:31 .. authoring or content management strata it might be useful to have multiple 15:47:41 .. languages in the same document. Frankly I don't think we've pursued many of 15:47:56 .. those thoughts much further since then, so this is bringing back those issues I think. 15:48:10 Pierre: I think a data point here is there is a difference between those things 15:48:22 .. that change dynamically during presentation. Language does not change, 15:48:28 q+ 15:48:31 .. but responsive design is dynamic changes during the presentation like 15:48:42 .. rotating the phone for example. That may be a differentiating point. There is 15:48:54 .. also a continuum of say aspect ratios whereas language is a discrete quantity. 15:49:00 .. There are differences across those use cases. 15:49:17 Glenn: I note some users find switching language mid-presentation to be useful 15:49:31 .. but they are fewer in number. If I'm watching a film with Russian and English 15:49:48 .. subtitles I might want to see Russian as a learner, but go back to the English 15:49:57 .. if I don't catch it all. Probably a corner use case! 15:50:07 Cyril: Glenn raised an interesting point about when to merge content. 15:50:21 .. For Netflix we receive deliveries of different languages from different vendors 15:50:35 .. at different times, so probably not. But forced narratives could be done in one 15:50:48 .. document. Certainly multiple aspect ratio could be. It makes sense when it is 15:50:58 .. the same content displayed differently. When it is different content we would 15:51:09 .. use a different document. The question is still open for forced narrative content. 15:51:11 ack c 15:51:27 Glenn: At one point in the history of TTML we had AFXP, the authoring profile. 15:51:41 .. When we were on a parallel path of defining them separately we recognised that 15:51:57 .. for AFXP it would be appropriate to have multiple languages but for DFXP 15:52:09 .. for distribution it would make sense to split them up. We haven't really 15:52:20 .. talked about authoring vs distribution profiling before but it may come back again! 15:53:23 Nigel: That's a really interesting discussion, thank you. 15:53:38 .. If there are no more requirements to discuss, I'll move us on in the agenda. 15:53:54 Topic: AOB 15:54:07 Nigel: Just one item from me, which is next week's meeting. Are there any 15:54:23 .. advanced regrets? Gary has already sent his. 15:54:30 group: [silence] 15:54:45 Nigel: Okay just confirming next week's meeting will go ahead and after that, 15:55:02 .. our next scheduled meeting is 10th January. Is that okay for everyone? 15:55:16 group: [silence] 15:55:42 Pierre: Okay with me. 15:56:03 Nigel: And of course a reminder we want all requirements to be raised as issues 15:56:28 .. by the 20th December which is the date of next week's call. I don't think that 15:56:36 .. means they have to be opened by the beginning of the call next week, but it 15:56:40 .. would possibly be helpful. 15:56:58 .. The main driver for that was to allow for adequate review time over the break. 15:57:45 Topic: Meeting close 15:58:10 Nigel: Thanks everyone, meet again same time next week. Bye! [adjourns meeting] 15:58:13 rrsagent, make minutes 15:58:13 I have made the request to generate https://www.w3.org/2018/12/13-tt-minutes.html nigel 16:01:34 s/to address to I/to address so I 16:01:58 s/until a later date if we need a/until a later date, if we need a 16:03:24 s/.. closed from the review stage/Nigel: closed from the review stage 16:04:18 s/.. new things in TTML, just clarifications./Cyril: new things in TTML, just clarifications. 16:07:19 rrsagent, make minutes 16:07:19 I have made the request to generate https://www.w3.org/2018/12/13-tt-minutes.html nigel 16:09:22 scribeOptions: -final -noEmbedDiagnostics 16:09:23 rrsagent, make minutes 16:09:23 I have made the request to generate https://www.w3.org/2018/12/13-tt-minutes.html nigel 17:53:57 Zakim has left #tt