[Asterisk-Users] Tweaking AGGRESSIVE_SUPPRESSOR

[Dennis Webb]

Thanks for both of the responses.  I have 2 questions now, is there a
better codec to use?  I know if I can limit the voip delay that exists,
it might help the problem some.  20ms is what is current but maybe 10
would help.  What would be best, bandwidth is not an issue currently. 
This doesn't fix the wctdm issue obviously, which makes perfectly good
sense, but might would help.

Second question, what about a chennel bank?  I here people talking about
the adit 600 but from what I read, it seems more like it requires a T1
as it's a DSU if I'm correct.  This would bypass the tdm boards and
everybody says it has great echo cancellation.  I'm recompiling the
drivers right now and turning on the MMX optimizations hoping it might
help the issue (a man can dream can't he).  I'm getting desperate here
because the system works great except for the echo which the users are
getting irritable with it and I agree with them they should.

Once again thanks.  We've been running full steam with asterisk for over
2 months now.

On Mon, 2005-03-07 at 18:14, Steve Underwood wrote:

> Steve Kann wrote:
> 
> > What he describes is echo suppression.  Because an echo canceller can, 
> > generally, only remove some part of an echo, not the entire echo, 
> > systems are generally designed to suppress the residual echo in some 
> > circumstances.  Old speakerphones had poor on no echo cancellation, so 
> > the suppression kicked in like that, because it was the only choice.  
> > In modern systems, the echo cancellation is much better, so 
> > suppression is not needed as much, and when it is used, it's probably 
> > done much more imperceptibly (with comfort-noise and stuff like this).
> 
> Only a very few high end conferencing speakerphones have ever used echo 
> cancellation. Even most expensive digital phones on PBXs merely do echo 
> suppression in speakerphone mode.
> 
> The nature of A-law/u-law limits the performance of an echo canceller 
> across the PSTN to about 30dB of echo improvement. If you look at the 
> behaviour of those codecs, you will see they give a roughly contant 30dB 
> of instantaneous dynamic range, and the echo cancellation enhancement 
> will never exceed that dynamic range. There is still enough residual 
> echo that good quality cancellers have to perform non-linear suppression 
> to eliminate it, and substitute comfort noise. 30dB, on top of the 
> minimum of at least 12-15dB of echo suppression the hybrids give, means 
> the echo should be rather quiet. It is still enough to annoy people, 
> though, and suppression is standard practice. It is specified in G.168.
> 
> Regards,
> Steve
> 
> _______________________________________________
> Asterisk-Users mailing list
> Asterisk-Users at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-users
> To UNSUBSCRIBE or update options visit:
>    http://lists.digium.com/mailman/listinfo/asterisk-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.digium.com/pipermail/asterisk-users/attachments/20050307/0961c2b5/attachment.htm