|
Kind of, I think the various ec algorithms train differently, but they end up with a
list of numbers which represent how much of the primary sound to replay (inverted
of course) and when, so in my case the ec needs to take
the current sample, and subtract the value of the sample 3.5ms ago to cancel
out the echo. The actual algorithm looks a bit like: Sn = current sample Sn-1, Sn-128 = previous samples E1-E128 are the
recorded coefficients S0 = S0 + Sn-1 * E1 + Sn-2 * E2 + … Sn-128
* E128 But anyone else feel
free to step in to correct me! Thanks James From:
[EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Eric Bishop Wouldn't that large spike
be the primary sound rather than the echo? On 1/15/06, James
Harper <[EMAIL PROTECTED]
> wrote: Here's the graph of the echo coefficients I grabbed from a x100p
card on my asterisk server. If my interpretation is correct, it shows that most
of the echo comes in at about the 28th tap, and assuming a sample
rate of 8000hz, that would be about 3.5ms. Will that tell you the sort of things you need to know? Thanks james From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]
On Behalf Of Eric Bishop Yes, plese do post it.... On
1/15/06, James Harper <[EMAIL PROTECTED]>
wrote: I'm working on some code to be able to preload the echo
cancellers (and obviously dump the data too, which I've already done). I should
have a patch ready tonight or tomorrow. If you are interested I can attach a plot of the coefficients
against time which might tell you the sort of thing you need to know if you
could do it on your data. The image is only an 11k gif… would it be okay to
post it to the list? James From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]
On Behalf Of Eric Bishop Does
anyone know how to determine the echo tail size (in ms) of a particular call?
|
_______________________________________________ --Bandwidth and Colocation provided by Easynews.com --
Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
