[PATCH 1/2] Comments in speechd.conf.in updated and polished

[Hynek Hanke]

On 24.8.2010 16:00, Chris Brannon wrote:
> The advantage of using something like libao as the sole audio backend is
> that less code is needed.  Non-existent code doesn't need to be
> maintained.  Someone else has already dealt with the eccentricities of
> the various audio backends, so why not take advantage of that?  It's a
> fairly small library, and you can build it to support only the backends
> that are actually used on a given system.
>    

I agree. Developing our own audio subsystem has been
forced by circumstances, but if libao now fulfills all our
needs, we should use it instead. That would be a big
simplification.

>    Also, I notice a delay when trying to stop speech
> with the libao driver.  I'm not sure what causes it.
>    

That sounds like something connected with buffering
settings. From what I've seen, our latency requirements
are quite special compared to other applications in the
desktop world, so I think the important think is whether
libao allows us to negotiate these requirements with the
target audio system (like the various buffering parameters
in Pulse Audio etc.).

Best regards,
Hynek