Re: SMIL/recommend.html (was [w3c-wai-ua] <none>]

At 15:31 1998/07/06 -0500, James Allan wrote:
>An action item from the July 2 teleconference was the formation of a
>subcommittee to look at SMIL issues.
>As a first step, participants should review the following document, for UA
>implications and issues.
>Jim Allan, Statewide Technical Support Specialist

That document 
    *DRAFT* SMIL accessibility recommendations
needs a date and a contact.

I hope that rate control would be dynamic, user adjustable during
listening in private. 

Concurrent source sharing is problematic. Different users with 
different reading speed controls would soon get out-of-sync if 
they were trying to share a common video or supplemental HTML or 
other scripted graphic source concurrently.

Internationalization needs mention. Are there ever needs for simultaneous
synchronized playback of captioning in different languages? In different
windows, through different audio channels.

The average text lengths for different languages differ, both because
of different language issues of wordiness and redundancy, and the
different translator's skills. With audio description, different scene
and environment choices would undoubtedly be made for independent

Speech rate needn't be proportional to fundamental voice frequency. 
If word boundaries and punctuation were known, those allow ready 
places for changing pause duration. So, it seems that a fairly 
wide range of speech may be tolerable, without doing violence to the
individual word intelligibility, the characteristic frequency of
the individual speakers, and the inflections they use. 

I believe voice compression/stretching first chops or lengthens the
vowel sounds and pauses. Consonant sssttrrrreeeeettccchhh sounds
unnatural. Vowel streeeetch is more reasonable. I posit that 
inflections may suffer as the duration and frequency change rates 
would be less recognizable the further they deviate from normal. 
They remain useful, as replacement for visual punctuation, and for

I hope that either voice or keying can be used to initiate searches
on a textual version, with a pronouncing aid to allow ambiguity
resolution such as "to, two, too" but not "tutu!"

Regards/Harvey Bingham
Yuri Rubinsky Insight Foundation

Received on Tuesday, 7 July 1998 00:25:30 UTC