I've spent the last week or so troubleshooting echo problems at my Wife's business, and I've been able to clear up about 99% of the echo, but there is still a little residual echo that I can't seem to "tweak out". The users describe it as "buzzing or crackling", but what it sounds like to me is a slight echo, but just of one syllable of the word. I've followed the numerous suggestions in the mailing list archives which is what has enabled me to get this far. After trying all the echo cancelers, and all the settings on each I settled on: - KB1 (with AGGRESSIVE_SUPRESSOR) - echocancel=128 - echotraining=600 - rxgain=0 - txgain=0 If I turned up the echocancel to 256 I'm able to eliminate the "buzzing" echo, however, on inbound calls I get a HUGE very loud echo that takes a good 20 seconds to go away (which I don't get on echocancel=128). Does anyone have any suggestions on how troubleshoot/tune this to eliminate the "buzz"? Here's my setup: Asterisk 1.2.9.1 Zaptel 1.2.6 TDM400P (3xFXO, 1xFXS) 6 Cisco 7960's running chan_sccp2 Along the same lines, in my hunt for echo, I was rather disappointed that I couldn't really find any tools to help troubleshoot the issue. It seemed like the general consensus was "Play with all the settings until it sounds good". I did manage to compile the Zaptel driver with the Zaptel "preload" patch, which helped me visualize my echo problems a bit, but preloading didn't get rid of the buzz. Also, I've eliminated inside wiring by moving the Asterisk box to my demarc to test, I also tried the TDM400P on different hardware to ensure it's not a motherboard issue. But, at the end of the day, those are all just stabs in the dark... If anyone has any suggestions, I'd sure appreciate it! Thanks! Brian
On 6/8/06, Brian Swan <swannie@swannie.net> wrote: [snip]> > I've followed the numerous suggestions in the mailing list archives > which is what has enabled me to get this far. After trying all the > echo cancelers, and all the settings on each I settled on: > > - KB1 (with AGGRESSIVE_SUPRESSOR) > - echocancel=128 > - echotraining=600 > - rxgain=0 > - txgain=0 >[snip] My personal preference, which works best in the most places is: - MG2 (with default DEFINEs, no aggressive etc) - echocancel=yes - echotraining=yes - rxgain=0 - txgain=0 Although in some cases: - rxgain=-0.3 - txgain=-0.3 seems to help a tiny bit. The MG2 canceller made a HUGE difference to our success rate in removing echo, I do not rate the updates to MG2 that are in trunk, but the released version (1.2.6) is excellent.
Brian Swan wrote:> I've spent the last week or so troubleshooting echo problems at my > Wife's business, and I've been able to clear up about 99% of the echo, > but there is still a little residual echo that I can't seem to "tweak > out". The users describe it as "buzzing or crackling", but what it > sounds like to me is a slight echo, but just of one syllable of the word. > > I've followed the numerous suggestions in the mailing list archives > which is what has enabled me to get this far. After trying all the echo > cancelers, and all the settings on each I settled on:...> If anyone has any suggestions, I'd sure appreciate it!Consider getting a Sangoma A200D (http://www.sangoma.com/datasheets/p_a200-specs) with the optional hardware echo canceller module. It "just works" for echo cancellation; no tweaks required. It takes a while to figure out how to install it, but once it's working it's great! - Mike
More taps in the Zaptel echo can = higher CPU usage + longer training times. In my experience, specifying an overly high number of taps (even 256) will bog the system when it is under load and lead to subjective delays in the audio path. But if you are going to be only doing 1 or 2 concurrent calls, why not experiment? -----Original Message----- From: Brian Swan [mailto:swannie@swannie.net] Sent: Friday, June 09, 2006 11:29 AM To: Asterisk Users Mailing List - Non-Commercial Discussion Subject: Re: [Asterisk-Users] Fun with Echo My wife is the customer, so she tends to give me funny looks when I ask her to sign waivers. ;) In all seriousness, though, I do recall reading that the 256 tap limit is "enforced" in software and that you can add/remove/whatever a few lines of code, and go up to 1024 (or greater I'd assume). Has anyone tried this? I'm considering attempting it this weekend to see if it does anything. Does more taps always = better echo cancelation? Thanks! Brian On Jun 9, 2006, at 12:15 PM, Mike Fedyk wrote:> Steve Davies wrote: >> On 6/9/06, Dr. Michael J. Chudobiak <mjc@avtechpulse.com> wrote: >>> >>> Consider getting a Sangoma A200D >>> (http://www.sangoma.com/datasheets/p_a200-specs) with the optional >>> hardware echo canceller module. It "just works" for echo >>> cancellation; >>> no tweaks required. It takes a while to figure out how to install >>> it, >>> but once it's working it's great! >>> >> >> Yes, I must agree that hardware echo cancellation from Sangoma (and I >> am sure Digium and others) is excellent, but it does add to the cost >> quite significantly sometimes. > If you haven't had your client sign a waiver that there maybe echo > because of the cheap hardware, then don't use the tdm400p. I've > wondered why the software echo can in zaptel doesn't go up to 1024 > taps and causes major problems at 256 taps. The tdm400p cards > don't have a high port density, you'd run out of slots before > hitting CPU barriers so why not use some of those resources for an > echo can with a longer tail? > > Use either the sangoma cards with echo can, or get a tdm2400p with > echo can. You will have happier customers if you do. > _______________________________________________ > --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_______________________________________________ --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
Brian, You may actually get better results lowering size of the buffer. Here are my settings that have been working nice so far on TDM400 (1+ year): echocancel=64 echotraining=800 echocancelwhenbridged=yes rxgain=2.0 txgain=-4.0 Notice difference between receive and transmit levels. That's essential, too. Andrei (MPI) Brian Swan wrote:> I've spent the last week or so troubleshooting echo problems at my > Wife's business, and I've been able to clear up about 99% of the echo, > but there is still a little residual echo that I can't seem to "tweak > out". The users describe it as "buzzing or crackling", but what it > sounds like to me is a slight echo, but just of one syllable of the word. > > I've followed the numerous suggestions in the mailing list archives > which is what has enabled me to get this far. After trying all the > echo cancelers, and all the settings on each I settled on: > > - KB1 (with AGGRESSIVE_SUPRESSOR) > - echocancel=128 > - echotraining=600 > - rxgain=0 > - txgain=0 > > If I turned up the echocancel to 256 I'm able to eliminate the > "buzzing" echo, however, on inbound calls I get a HUGE very loud echo > that takes a good 20 seconds to go away (which I don't get on > echocancel=128). Does anyone have any suggestions on how > troubleshoot/tune this to eliminate the "buzz"? > > Here's my setup: > > Asterisk 1.2.9.1 > Zaptel 1.2.6 > TDM400P (3xFXO, 1xFXS) > 6 Cisco 7960's running chan_sccp2 > > Along the same lines, in my hunt for echo, I was rather disappointed > that I couldn't really find any tools to help troubleshoot the issue. > It seemed like the general consensus was "Play with all the settings > until it sounds good". I did manage to compile the Zaptel driver with > the Zaptel "preload" patch, which helped me visualize my echo problems > a bit, but preloading didn't get rid of the buzz. Also, I've > eliminated inside wiring by moving the Asterisk box to my demarc to > test, I also tried the TDM400P on different hardware to ensure it's > not a motherboard issue. But, at the end of the day, those are all > just stabs in the dark... > > If anyone has any suggestions, I'd sure appreciate it! > > Thanks! > Brian > _______________________________________________ > --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 > >
I wrestled and wrestled with echo issues for months before finally just following the guide here: http://www.voip-info.org/wiki/view/Tellabs+Hardware+Echo+Cancellers And with help from the list I was able to COMPLETELY remove all echo for under $100.. Requires a tiny bit of simple soldering, but it works like a charm. The software cans just never did the job well enough, no matter which one I picked. Brian Swan wrote:> I've spent the last week or so troubleshooting echo problems at my > Wife's business, and I've been able to clear up about 99% of the echo, > If anyone has any suggestions, I'd sure appreciate it! >Consider getting a Sangoma A200D
> There is a spec for echo cancellation on PSTN called g.168. I believe > it's a > suite of tests which put the echo canceller through its paces and ifyou> pass > them you are certified to conform to g.168. None of the echocancellers in> zaptel conform to this, whereas the Octasic, Tellabs and otherhardware> echo > cancellers all do. If someone were to put the effort and energy into > making > the software echo cancellers compliant, you should find similarresults to> the hardware echo cans.Just out of curiosity, what form does the testing take? Is it simply a set of audio sample pairs (outbound, and inbound with some echo of outbound) that you run the ec against and the measure the results? What metrics are measured? I can only think of: . the amount of echo left after the ec has run . training time How hard would it be to make a g.168 test bed that the Asterisk ec's could be run against? I can definitely see the value in even just making some audio samples for people to listen to and giving each ec (with version number and tunable parameter values) a score against each sample, even if it's just a subjective score based on perception from users. That way a newbie could listen to the audio samples to find one that sounds like the sort of echo they are getting (mainly based on delay I guess but different types of echo will have different frequency responses than others) and then can pick a good ec based on score. Comments? James
yeah this post is old and there have been dozens of replies, but here's some feedback for the list, now that i have some. we're using a sangoma a102 card (no hw ec) with 2 pris from sbc. asterisk 1.2.7.1, zaptel 1.2.6 (much testing previously with 1.2.5). we first used: KB1 (not aggressive), echocancel yes (128), echotraining yes (400), rxgain/txgain 0. most calls with this setup were fine, but maybe 5% of pri calls were reported as having anywhere from slight to completely unbearable echo. playing with echocancel yielded worse results. 256 got results similar to Brian's report below. 64 yielded echo on calls from numbers that never had echo before. we're now pretty settled on similar settings to Brian's: KB1 (WITH aggressive), cancel yes (128), training yes (400), gains 0. AGGRESSIVE_SUPRESSOR made a huge difference for us. reports of bad echo on calls dropped from one or two reports per day to 0 since we started using this setup about 2.5 weeks ago. i still receive the occassional report of slight echo on a call or two (or slight echo at the beginning of a call that goes away), from people who i actively ask about it, but for the most part everyone is very content. i also played around with MG2 (non-aggressive), per suggestions on this thread, but aggressive KB1 showed better results in a few tests. i have not as of yet tried aggressive MG2. looking over our cacti logs, load average on the system didn't noticeably change between any of our echo handling changes. anyway, i know there's no silver bullet for software ec, and anyone reading this looking for ideas should be sure to keep that in mind, but this thread gave us a number of ideas to try, so i wanted to report some of the results. good luck, all! -tcl. On Thu, 8 Jun 2006, Brian Swan wrote:> I've spent the last week or so troubleshooting echo problems at my Wife's > business, and I've been able to clear up about 99% of the echo, but there is > still a little residual echo that I can't seem to "tweak out". The users > describe it as "buzzing or crackling", but what it sounds like to me is a > slight echo, but just of one syllable of the word. > > I've followed the numerous suggestions in the mailing list archives which is > what has enabled me to get this far. After trying all the echo cancelers, > and all the settings on each I settled on: > > - KB1 (with AGGRESSIVE_SUPRESSOR) > - echocancel=128 > - echotraining=600 > - rxgain=0 > - txgain=0 > > If I turned up the echocancel to 256 I'm able to eliminate the "buzzing" > echo, however, on inbound calls I get a HUGE very loud echo that takes a > good 20 seconds to go away (which I don't get on echocancel=128). Does > anyone have any suggestions on how troubleshoot/tune this to eliminate the > "buzz"? > > Here's my setup: > > Asterisk 1.2.9.1 > Zaptel 1.2.6 > TDM400P (3xFXO, 1xFXS) > 6 Cisco 7960's running chan_sccp2 > > Along the same lines, in my hunt for echo, I was rather disappointed that I > couldn't really find any tools to help troubleshoot the issue. It seemed > like the general consensus was "Play with all the settings until it sounds > good". I did manage to compile the Zaptel driver with the Zaptel "preload" > patch, which helped me visualize my echo problems a bit, but preloading > didn't get rid of the buzz. Also, I've eliminated inside wiring by moving > the Asterisk box to my demarc to test, I also tried the TDM400P on different > hardware to ensure it's not a motherboard issue. But, at the end of the > day, those are all just stabs in the dark... > > If anyone has any suggestions, I'd sure appreciate it! > > Thanks! > Brian > _______________________________________________ > --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 >