Most standard computers have speech synthesizers available these days.
Transfering text string
for local synthesis saves bandwith and pushes the computation out to the
clients.
I was considering using this in a question answering system. I would like
to return both html and
some text for the sythesizer to say on display of the html page.
A more elaborate idea would associate inline speech with positions in the
document such that
when they became visibile (via scrolling) they would be queued for synthesis.
If the mechanism provides a means to specify the rendering, one can perhaps
control the voices
via a generic mapping and one might use the same facility for non-speech
inline audio.
Where do we stand on this for HTML 3?