Re: [AD] audio input latency, changes &new func proposal

[Mike Fahlbusch <mcf@xxxxxxxxxx>]

At 01:39 AM 19-12-03, you wrote:
> At 01:33 PM 18/12/2003 +0100, you wrote:
> > > Q.  can the buffer sizes be changed so they return smaller buffers (low
> > > latency)  ?
> >
> > I think so, but you should experiment.
> >
> > > if the buffer size can be adjustable, can i propose a new function
> > > called  set_sound_input_buffer_size(size_t i);
> >
> > Yes, this makes sense.  But I think extending start_sound_input is better:
> >
> > int start_sound_input_ex(int rate, int bits, int stereo, int delay);
> >
> > with 'delay' in milliseconds.  The buffer length calculation would then be
> > done automatically.
>
> yes, this seems a very good plan..  ensures that the value is set before 
> start_sound_input() can return a value.
> i shall  investigate further where this code must go, i am still tracing 
> out the code map (what calls what; what vars are relavent etc..) 
> for  sound input.

Has anyone found a solution to the win32 sound recording latency 
problem?  I need to get smaller buffers (around 512 to 2k bytes or 30 to 
125ms ) for a speech recognition app.  It works fine in DOS and Linux, it's 
just win32 that I can only get 1s recording intervals.
I tried experimenting with different values in the allegro source but 
couldn't get it to work with smaller values.


Sig:
Think outside the dot.



Regards,
       Mike

Sig:
Think outside the dot.