[PATCH] Reopen libao device handles when audio parameters change.

[Trevor Saunders]

Hi,

On Wed, Sep 01, 2010 at 10:47:15AM +0200, Andrei Kholodnyi wrote:
> On Wed, Sep 1, 2010 at 10:16 AM, Halim Sahin <halim.sahin at t-online.de> 
> wrote:
> > Hi,
> > On Wed, Sep 01, 2010 at 01:09:08AM -0400, Trevor Saunders wrote:
> >> why? ?that makes no sense, we are asking the module to play a wave file.
> >> There's no reason I can see to involve the synth, if you see a reason to
> >> involve the synth please tell me what it is.
> >
> > +1.
> > Which synth can play soundicons except espeak?
> > I didn't use soundicons but if speechd wants to support it so it should
> > be done for all synths and handled by sd itself and not through the
> > synth.
> 
> +1, but in configurable and flexible way.

sure, configurability is a feature worth looking at adding.

> > Which synth can play soundicons except espeak?
> ibmtts, nuance, svox and loquendo.
> 
> and I do not see any reasons why we should avoid using those capabilities.

we aren't avoiding them so much as not going out of our way to use
them.

> Why do I need this extra stuff in SD if my synth can do this job for me?

because its already going to be there, but by not involving the synth we
can take stuff out of sd.

> we shall provide this functionality in SD for the synths that do not
> support it, no doubt about it.
> 
> >> yes, and this is imho very silly what will happen is that espeak will
> >> take that message, and just turn around and call the call back we gave
> >> espeak for <auido> tags. ?Then we implement a function like the <audio>
> >> callback for espeak in each five modules. ?Its the same basic function
> >> in five places, and there is other code in the code specific to modules
> >> to deal with sound icons, unless there is some good reason to involve
> >> the synth in sound icons that I can think of we can get rid of a lot of
> >> repeated code by never involving the synth in the sound icon msg type.
> > ACK.
> 
> NACK, it is not silly, since e.g. espeak also supports SSML.
> Do you want to parse SSML, extracting audio tags from it and play them in SD?

no, that would be going out of the way to not use the synth which is the
reverse of what is going on here.

I think a patch will make this clearer but what I'm thinking is roughly
this we'll need and infact have in several places a play_wave_file()
function not by that name, but thats what they do.  it's simply easier
to have do_sound_icon() just call that play_wave_file() function
directly than to make module_speak() worry about sound_icons.  Of course
for espeak we'll still have a handler for <audio> call backs, but we
won't generate messages that are just an audio tag just to call that
handler, and that handler will just get the file name from espeak and
call the play_wave_file() function.
I'll try to get a patch out tonight.

Trev

> 
> _______________________________________________
> Speechd mailing list
> Speechd at lists.freebsoft.org
> http://lists.freebsoft.org/mailman/listinfo/speechd
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: 
<http://lists.freebsoft.org/pipermail/speechd/attachments/20100901/d956405b/attachment.pgp>