[PATCH 1/7] move module_audio_init from each module to the module_util.c

[Andrei Kholodnyi]

On Tue, Nov 30, 2010 at 11:34 AM, Bohdan R. Rau <ethanak at polip.com> wrote:
> Personally I am a little lazy, but I'll try (I only don't know when).

This is good.
I believe we could clean this area first,
i.e. separate audio backends and modules + generic module modifications
and then switch to the points you have addressed below.

> One important question:
>
> I found old but working commercial Polish synthesizer - syntalk. I wrote
> simple draft (generic module for 0.6.x/0.7.x), one of my (blind) friend
> said it's rather usable. The problem is simple: syntalk does not support
> pitch and rate changes. I used play command with tempo and pitch
> parameters, but for me it should be good to use some internal
> speech-dispatcher function instead.
>
> In fact I wrote this function (as draft of course), but function needs
> sonic library for rate, and samplerate+sonic for pitch.
> So question is simple: as this function will be used in minimum two
> modules (ivona and generic), should it be placed as common audio function,
> or should it be private for these modules?

> I also try to implement SAPI module (I won't wait for OpenSapi to be
> usable, my friend started with writing windows part, if he will finish his
> work I need only some hours to write speech-dispatcher module), and I also
> want to use sonic and resampling... so may be it would be good idea to make
> pitch/rate changer as common audio function?

We have discussed a while ago that it might be a good idea to
integrate sonic lib into speechd.
I haven't seen any volunteers for this work so far :D

> And may be using samplerate is overkill... may be I should use simpler
> resampling algorithm directly in this function?

if it is already existing and working why shouldn't we use it?
The only reason I could see it is not part of the particular Linux distro.