lkml.org 
[lkml]   [1999]   [Sep]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Turning lucent winmodem into soundcard
Steve Underwood wrote:
> If you aim to have most output generated in user space, and some fudged
> underflow fill in generated in the kernel, how will you make the echo
> cancellor function reliably?

I'm still undecided about whether to fill with idle frames or not.
Alan's convinced me that we can try without, and if it causes real
problems on real connections, add it later. It might be fine.

Echo cancelling means the Tx side needs to know when an underrun occurs,
and precisely how many sample points were transmitted in the underrun.
Because, as you point out, it needs to know the exact signal that's
transmitted.

> If the output stream can underrun, its more likely to happen during
> the initial training period, as the machine is working harder.

During the training period, modem latency is not important so we can use
a larger output buffer and underruns won't happen.

> Just letting the echo cancellor hiccup during the transmit underrun
> might be a poor solution, as it may screw up convergence.

Assume we know exactly what's transmitted, even during an underrun.

Does the echo canceller adapt so quickly that the change in signal
character will screw it up?

If so, then would having the kernel driver fill with a
zero-autocorrelation pattern fix that?

> You could relegate the whole training process to
> the kernel, but that puts more bloat in the kernel.

That's the one to look at after all the others have failed.

> Remember V34 is fairly fussy about the characteristics of the A/D and D/A
> converters. A lot of sound cards may not function very well in this role.
> Weird phase effects in some converters really screw up the dynamic
> performance of the echo cancellor. The same phase effects don't have much
> effect on the actual demod process, as this dynamically adapts well to such
> effects. Still, I guess many sound cards will work OK.

You seem to know a lot about echo cancellation. Can you point me to
good, freely available references on the subject? I have a design, but
it's based on hand-waving and dodgy maths...

What sort of phase effects are present in sound card converters?

thanks,
-- Jamie

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:54    [W:0.039 / U:0.696 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site