The Problem: When a paragraph or other region contains live/volatile data that
changes frequently, the speech output might be out of date by the time the
system speaks that data (because the string is sent to the speech engine all at
once). One example of this is a countdown timer.
It would be nice if the system could read the updated information when the
speech gets to that point in the speech.
Such regions could be marked up with:
aria-role="timer" (see http://www.w3.org/WAI/PF/aria/roles#timer)
or
aria-live="*" (but supported as if it were aria-live="off"; see
http://www.w3.org/WAI/PF/aria-practices/#liveprops)
The string sent to the speech engine could be split on such live regions so
that the system could wait for some speech to get done, then check the live
value, speak it, and then speak the rest of the text.
The one downside is that the speech engine may not have the best intonation and
inflection because it is being sent only substrings, but I don't think there is
a good way around that.
Original issue reported on code.google.com by jbjor...@gmail.com on 21 Aug 2013 at 5:37
Original issue reported on code.google.com by
jbjor...@gmail.com
on 21 Aug 2013 at 5:37