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

[Hynek Hanke]

On 24.8.2010 15:20, Chris Brannon wrote:
>> -# AudioOutputMethod "pulse,alsa"
>> +# AudioOutputMethod "pulse"
>>      
> You might want to mention this one in the commit message, because it
> seems to be a significant change.
>    

I thought we have already fixed this in src/server/config.c
but now I see we didn't, so I'll post a separate patch for that.
Thanks for spotting this.

>>   #AudioPulseMaxLength -1
>>      
> The parser doesn't recognize this option anymore.
>    

I see, so apparently we need a more detailed revision
of the config file after the audio subsystem rework
rather than my improvements in the descriptions :)

> Same.  However, the parser does recognize AudioPulseMinLength, which is
> not documented in the config file.
> I wonder whether AudioPulseMinLength has the same meaning as
> AudioPulseMinRequest?
>    

Comment in src/audio/pulse.c:44 says:

/* This is the smallest audio sound we are expected to play immediately 
without buffering. */
/* Changed to define on config file. Default is the same. */

It is the value passed into Pulse Audio as tlength, which is
described as target length of the buffer.

http://0pointer.de/lennart/projects/pulseaudio/doxygen/structpa__buffer__attr.html

I'm not sure it really is what is described in the comment above. Furhter,
it's recommended to be set to the minimal acceptable latency, which seems
to me is quite above 100 bytes.  Minimum request is something different.

Maybe Ruy or somebody who participated in the PA audio module
rework could clarify how this works?

Thank you and Best regards,
Hynek Hanke