W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > January to March 2002

Re: Audio captioning - practical questions

Date: Fri, 15 Feb 2002 11:19:47 -0500
Message-Id: <sc6ceedb.092@HQF2.eeoc.gov>
To: <w3c-wai-ig@w3.org>
The reason we feature a Hispanic female speedskater, a skier who lost a leg to cancer, and an African-American who heads the USOC and has been a Fortune 500 CEO, is as examples of people who are taking full advantage of their "Freedom to Compete" (the theme of the PSA campaign) and have achieved, or expect to achieve, great things. So who they are matters, but what they're doing in the video isn't significant.

Some of the technical issues you raise (why not use SMIL to select bitrate, language, etc.) are interesting. As far as picking language is concerned, I want people to be able to see/hear either language without having to change their player settings. That eliminates any solution based on testing for language, and unfortunately adds to the number of links. It's awkward, but this isn't a web campaign, it's intended as information about our TV campaign - a subtle difference that significantly shapes our approach.

Testing for bitrate is what I had originally wanted to do, instead of using separate links for each of three different versions. But in reading through the documentation from Real, there appears to be a serious limitation. My two low-end versions are small in terms of screen size, the high-speed version is twice the height and width. Although Real's implementation of SMIL allows me to switch among the three options, and will automatically resize the video window, there's nothing I can find, and nothing the Real tech support people can find for me, that allows me to modify the layout of the root window based on bitrate. So a large window with a neatly centered high-bitrate video means an oversized window with a misplaced, off-center low-bitrate video. Not critical, but ugly, and my bosses wouldn't be happy. I'd have to redesign it to accommodate both video sizes, and I didn't have the time. If anyone knows of a solution to this, please let me know. (I assume it can be done in SMIL, just not in Real's implemetation of SMIL.)

Testing for system-captions is probably a much better approach than I used - I'll probably end up redoing that.

As far as sneaking in a quick word or two in the gaps in the audio, that's probably all I'll be able to do. I think you're right about not doing too much preloading. But that limits the amount of information I'm able to include. Which raises the more difficult question of how could this be accomplished in a more complex video where there is a greater need for audio descriptions but no more room in the audio track? As an example, here's another EEOC video clip:

http://www.eeoc.gov/35th/videos/longintro-medium.ram (56k modem)
http://www.eeoc.gov/35th/videos/longintro-high.ram (broadband)

Thanks for the NCAM example - interesting (FYI - the instructions for displaying the captions don't work with RealOne). I think I'll need to spend some time on the NCAM site.

>>> Andrew Kirkpatrick <andrew_kirkpatrick@wgbh.org> 02/14/02 09:24PM >>>
What you have up looks great.    

>Considering a very short time frame, limited resources, and the fact that I've never worked with SMIL 
before, I think I did a decent job. What I don't have, however, is audio captioning.

I think you mean audio description (also called video description).

>How important is audio captioning for something like this? The video portion contains scenes of the two 
athletes in training, some close-ups, and head shots of the Olympic Committee CEO. They're visually 
appealing, but don't convey any real information. It's all eye-candy, while all of the "substance" of the PSAs 
is included in the audio, and the text captioning. Audio captioning is, if possible, desirable, but is it 

So what is the point of having this piece be about speed skaters at all?  It isn't actually mentioned in the 
audio - does it make a difference?  Or is it thas it is a female skater ripping along the ice that matters for this 
message? You have very little time for description, so you won't say much anyway.  An interesting question 
is how does an individual determine how much is necessary to describe -- section 508 does mention audio 
description (1194.4 definitions: Alternate methods. Different means of providing information, including 
...captioning, text-to-speech synthesis, and audio description.).  To me, it is the same question as is often 
asked regarding alt text, but more complex. 

>Assuming the answer to the first is "yes," how is this best accomplished? The technical side is easy - I can 
record a second audio track and use SMIL to sync it up with the main video stream. The problem is, I have 
short videos with lots of quick edits and scene changes, and a primary audio track that already contains 
music and voiceovers. There's not much room for additional description. How much information should be 
included? Is it OK to "pre-load" it with description at the beginning, before the video starts? What other 
methods are there of including audio captioning in an already crowded audio track? Any ideas or pointers 
to resources would be appreciated.

AWK: Ideally you sneak quick descriptions into the gaps in the audio program.  You could preload a little, 
but probably don't want to add more than a second or two.  I'll ask some of my colleagues who do this for a 
living to get their take.  You could, were it merited (and if real supported it) use extended descriptions, 
which can pause the timeline until the interrupting element is finished.  The Real One player is supposed to 
support this, but not all of the bugs are worked out yet.  In any case, I don't believe that approach is 
recommended here.  

>The videos as they exist now are available at: http://www.eeoc.gov/psa/index.html 
>Please take a look (or listen) and let me know what you think. A reasonably recent version of RealPlayer 
is required.

Another thing you might want to consider is using some of the smil test attributes to reduce the number of 
links on the page.  Refer to both sets of captions in the same smil file, with tests for language.  You can 
also separate the audio from the video and refer to both program audios, also with tests for language.  The 
captions would use the system-captions (systemCaptions) test attribute, and would play only if the user had 
the preference set in Real (you could provide information.  Come to think of it, you can also use the bitrate 
test attribute so the user would select the version based on their player preference for connection speed.  
So that gets you down to one smil file, and one link.  

If you add audio description, you will need another link, since RealOne is Real's only player that Real has 
that has a functioning preference for audio description.  Oratrix's GRiNS also plays Real media and does 
test for audio description (and does extended descriptions too!), but is not widely used.  So you would have 
a second link that plays the audio, the video, and the descriptions.  If you wanted to be really nice, you 
could make a third link with description and program audio only, thereby reducing the download for the end 

For an example that is similar, check out: http://ncam.wgbh.org/richmedia/media/carreal/carreal.html.  

Hope this helps,

Andrew Kirkpatrick, Technical Project Coordinator
CPB/WGBH National Center for Accessible Media
125 Western Ave.
Boston, MA  02134
E-mail: andrew_kirkpatrick@wgbh.org 
Web site: ncam.wgbh.org

-------- End of forwarded message --------

Andrew Kirkpatrick, Technical Project Coordinator
CPB/WGBH National Center for Accessible Media
125 Western Ave.
Boston, MA  02134
E-mail: andrew_kirkpatrick@wgbh.org 
Web site: ncam.wgbh.org
Received on Friday, 15 February 2002 11:20:01 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 20:36:07 UTC