Re: Embedding the track kind in WebVTT files

Yes, this is a good idea.  We've suggested it before.  Not least there is the case where the captioning is done by a  separate 'house' from the HTML5, and the HTML authors want some clue as to what's in the file.  I agree, they should be able to over-ride.  Self-describing files are a Good Thing, I think.


On Sep 12, 2013, at 2:30 , Cyril Concolato <cyril.concolato@telecom-paristech.fr> wrote:

> Hi all,
> 
> WebVTT files are/will be used in environments where there might not be an HTML page pointing to it. Some examples are:
> - WebVTT content embedded and multiplexed with video and audio in an MP4 file
> - Plain text WebVTT files referenced by an adaptive streaming manifest (DASH MPD, HLS playlist, ...)
> 
> All embedding environments typically enable marking a track with a language (e.g. an audio track) but not all can carry the kind types (subtitles, captions, metadata). In order to avoid losing that information in those environments, I think the WebVTT file should declare its kind. Also, when producing WebVTT files, it should be possible to indicate the language, kind or even label in the file itself, to avoid mistakes.
> 
> I would like to propose to define standard WebVTT metadata headers, as follows:
> - WebVTT metadata header name: kind
> - WebVTT metadata header value: as defined by kind attribute of the track element in the HTML 5 specification
> 
> - WebVTT metadata header name: language
> - WebVTT metadata header value: as defined by srclang attribute of the track element in the HTML 5 specification
> 
> - WebVTT metadata header name: label
> - WebVTT metadata header value: as defined by label attribute of the track element in the HTML 5 specification
> 
> Note that some people are already using non standard name/values for that (e.g. http://wiki.webmproject.org/webm-metadata/temporal-metadata/webvtt-metadata)
> 
> Of course, this introduce redundancy when the embedding environment also sets those attributes and might be problematic if the values differ. Resolving conflicts should be specified by the embedding environment. In HTML, the information in the HTML page should probably have precedence, to be able to use a caption track as metadata, if the HTML author wants to.
> 
> Comments ?
> 
> Regards,
> Cyril
> 
> -- 
> Cyril Concolato
> Maître de Conférences/Associate Professor
> Groupe Multimedia/Multimedia Group
> Telecom ParisTech
> 46 rue Barrault
> 75 013 Paris, France
> http://concolato.wp.mines-telecom.fr/
> 
> 

David Singer
Multimedia and Software Standards, Apple Inc.

Received on Thursday, 12 September 2013 17:11:44 UTC