- From: Henri Sivonen <hsivonen@iki.fi>
- Date: Mon, 17 Sep 2007 12:36:32 +0300
- To: HTMLWG WG <public-html@w3.org>
<video> accessibility was discussed on IRC today. Here are some issues that I think the spec should address. The spec already says that the fallback is not for old browsers. I think it should drive the point even more explicitly and say it isn't for accessibility alternatives in <video>-aware UAs and that accessibility should be handled in the video itself. Currently, the spec covers accessibility for the deaf (closed captioning). However, it doesn't cover descriptions for the blind. Closed audio descriptions for the blind could be handled using an additional sound track. However, whereas captions are negligible in data size compared to the video track, additional soundtracks may be large enough to make content providers hesitate to always serve them. Points that seem to need research: * Can Speex compress audio descriptions enough to be negligible compared to the video track so that content providers could be comfortable with always muxing in the audio description? * If not, how to handle audio descriptions as an optional HTTP download? Should there be a separate audio file or an entire alternative video file? * Would it be feasible to encode the descriptions as timed text and feed them to a speech synthetizer on the client? (Does any off-the- shelf timed media framework even support this sort of thing?) For closed captioning, it would be good to have at least a note pointing to information on how to flag an Ogg Writ track as closed captions so that authors and UAs are more likely to get the same understanding of what constitutes closed captions. (This may have to wait until later in the development of HTML5 and Writ.) Likewise for closed audio descriptions muxed in the main file, it would be good to have a note on how to flag an audio track as being a closed description track. Flash appears to make both closed captioning and audio description possible, so the above wouldn't be about pushing the envelope but merely about getting feature parity. -- Henri Sivonen hsivonen@iki.fi http://hsivonen.iki.fi/
Received on Monday, 17 September 2007 09:36:51 UTC