Re: [Alsa-user] driver approach

[Clemens L. <cl...@la...>]

"Alexander Carôt" wrote:
> > No, it isn't.  When your applications gets woken up by the interrupt
> > at the end of a period, it can handle all the data in that period.
> > Data that is 'older' than one period (i.e., before the previous period
> > boundary) was already handled by your application when it got the
> > previous interrupt.
>
> Alfons just replied the same way which confuses me. I don't doubt you're
> right but I want to explain what confuses me :
>
> 32kHz,128 Samples/Frame - The delay between the actual finger snap (recorded
> directly to the left channel of my recording device) and the play back from
> the ALSA-driven-soundcard recorded to the right channel is 8ms. So is it
> that it takes 4ms to capture and 4ms to play it back

Yes.  It's at most 4 milliseconds until the captured data is available
for your application, but similar considerations apply to playback --
when your application writes one period at a time to the playback
buffer, the samples at the end of the period will be output at least
four milliseconds after they were written.


HTH
Clemens