Re: GNOME "Telephony Application Programming Services" Proposed

[Bill Haneman <bill haneman sun com>]

On Thu, 2003-02-06 at 20:28, mcnichol austin ibm com wrote:
> Hello,
> 
> I've been following this thread and was wondering if people
> were thinking about using MAS (www.mediaapplicationserver.net)
> as a framework.
> 
> A few of us were batting some ideas around a while back, and we
> thought one solution for tts would be to have gnome-speech
> generate Speech Synthesis Markup Language (http://www.w3.org/TR/speech-synthesis/)
> which it would pass to MAS.  MAS could take care of
> delivering the data across the network (if needed).  On the
> desktop system MAS could load a plugin to convert SSML to speech, and
> finally deliver that data to the audio driver.  Something
> similar could also be done for braille.

Well, I would not use MAS as the primary transport layer here (between
gnome-speech and the TTS engine), but it could play an important role in
other places in the stack.

Note that gnome-speech already provides network-transparent
text-to-speech services (though the discovery protocol which it uses for
initial queries doesn't yet have a supported remote-query protocol). 
Once a service is detected it can be used remotely or locally in a
transparent fashion.  Also, upcoming versions of gnome-speech have plans
to support SSML markup.  So gnome-speech itself already has a roadmap
for markup support and non-coresident client-server scenarios.

For obvious bandwidth reasons there are cases where delivering the
actual audio over the network is undesirable, but of course there are
situations where it could be desirable to generate (or process) audio
remote from the desktop.  I think that it's in such cases that MAS shows
the most promise, since it can substantially assist with information
about latency, etc. and therefore support smarter
client-callback/notification mechanisms in high-latency scenarios.  A
more common case might be an application server scenario, in which the
application client of gnome-speech is remote to the user's desktop, and
the gnome-speech service instance is local (to minimize bandwidth
consumption).

Of course gnome-speech is agnostic about its back-ends, so a generic
SSML backend could probably be created, provided there was sufficient
support for callback/notification tags.

- Bill

> 
> Dan
> _______________________________________________
> gnome-accessibility-list mailing list
> gnome-accessibility-list gnome org
> http://mail.gnome.org/mailman/listinfo/gnome-accessibility-list
-- 
Bill Haneman <bill haneman sun com>