Re: [orca-list] [g-a-devel] gnome-speech, and audio output, moving forward.

[Michael Whapples <mwhapples aim com>]

My oppinion on this is that speech-dispatcher is very much the way
forward for a common speech API. Although Linux is all about choice,
there are sometimes cases where it would be better to have one common
system, and speech/audio is probably a good example of this. With the
choice of different standards, different developers choose one, but this
then means that a user may have two apps that use the two different
systems, so defeating one of the reasons for having a common API (that
the API should handle cases when two apps want the one resource).

So why would I say speech-dispatcher rather than gnome-speech? The first
reason, possibly not important to some users, but there are some of us
who use some native console stuff, and so need speech-dispatcher as it
isn't a gnome component, and is used by some non-gnome apps, where as
gnome-speech requires gnome and for those users who may not want gnome,
they may be forced to have it due to needing speech output if
gnome-speech became the standard. Another reason is that
speech-dispatcher, for those synths where there is a specific
speech-dispatcher module, the audio output is handled by
speech-dispatcher, so ALSA output can be used with festival, IBMTTS
(eloquence or viavoice are some other names for that synth), espeak and
I think flite, which resolves issues with multiple sounds being played
for those without sound cards supporting hardware mixing.
Speech-dispatcher has bindings in a few languages, java, python, C, and
Clisp, so making it available to a number of developers. I think Steve
mentioned about internationalisation, I think speech-dispatcher even has
support for that.

I think Luke once mentioned that ubuntu wasn't going to use
speech-dispatcher because it didn't offer sufficient client/user
specific configuration, particularly mentioning audio output selection.
As a response to that, I don't know how gnome-speech really provides
this as gnome-speech doesn't actually handle the audio from the synth,
so I thought the client is stuck on changing audio output with
gnome-speech. May be speech-dispatcher could be improved in this area,
IE. move audio output from module configuration to client configuration,
and for those modules that produce their own audio (generic based
modules), speech-dispatcher may be should provide a basic player which
puts the file through the speech-dispatcher audio system (eg. the synth
writes to a file, and then this file played by the SD player, so in the
case of the old generic based espeak module the aplay command might be
substituted by the sd player command).

Just my thoughts and recommendations.

From
Michael Whapples