Patch (rather for discussion): sonic and samplerate implementation

[Andrei Kholodnyi]

On Mon, Dec 6, 2010 at 10:39 AM, Bohdan R. Rau <ethanak at polip.com> wrote:
>> Since Bill has proposed to integrate libsonic,
>> I'd say the first step would be to grab sonic src and put it under
>> e.g. src/libs/sonic
>
> I don't think it's good idea.
> For me: libraries which may be used out of speech-dispatcher should be
> shipped separately. But of course it should be possible to build these
> libraries together with speech dispatcher, or must be accessible from
> speech-dispatcher site.
> There shoud be dotconf, sonic and dumbtts (I think).

Very good idea. However I'm neither Bill, nor William :D
Concerning dumbtts I guess the answer is clear. BTW, I have a couple
of patches for it.

>> Sonic functionality shall be generic to all modules, i.e. I do not
>> expect any modifications in ivona.c,
>> but rather in module_utils.c.
>
> It was necessary to add some code to ivona.c (like UPDATE_PARAMETER).

why? since it does nothing, why do you need it in ivona.c?

>> For those modules that do not provide a rate change it shall be called
>> constantly
>> and for the modules which supports it, it shall be some criteria when
>> this function shall be called,
>
> More: for some syntesizers where choices are "too slow" or "too fast"
> sonic may be used for precised rate.

yes, actually the problem with -100...0...100 that it does not tell
you for the different modules whether
100 is 3X or just 1.5X. And here sonic can help and normalize 100 for
all e.g. to the 3X.

>> Function spd_audio_effect you have implemented should probably pass
>> AudioTrack instead of samples.
>
> You mean somethink like:
> int spd_audio_effect(AudioTrack *track,...)

yes, something like
AudioTrack * spd_audio_effect(AudioTrack *track,...)
since you need an output as well.