[asterisk-users] MeetMe echo problems with more than two participants

[Matthew J. Roth]

Alessandro Russo wrote:
>
> we are using Asterisk 1.4.18.1 <http://1.4.18.1/> on debian 4.0 etch, 
> pwlib 1.10 and openh323 1.18.
>
> We are using MeetMe for conference calls and with two participants 
> there is no echo problems, but with more than two participants there 
> is a lot of echo that sometimes disappear for a short time and all 
> function well.
>
> Someone have some suggestions??
>
> Do you ever used app_conference 
> http://sourceforge.net/projects/appconference/  ??
>

Alessandro,

Are you certain that the echo isn't being introduced by someone on the 
conference using a speakerphone?  This would cause what is known as 
acoustic echo 
<http://en.wikipedia.org/wiki/Echo_cancellation#Acoustic_echo> and it's 
always my first suspect in a situation like the one you are describing.

This is not a problem that is specific to Asterisk and I'm fairly 
certain there is nothing that can be done within your configuration to 
correct it.  Instructing the conference participants to mute their 
phones when they aren't speaking or to use their handsets should reduce 
acoustic echo.  Some phones 
<http://www.voip-info.org/wiki/view/Uni-Ta+Technology> also claim to 
have a "full-duplex speakerphone with advanced acoustic echo 
cancellation," but caveat emptor.

That said, I'm not an expert on echo cancellation and I have an 
installation where the users are making similar complaints about echo 
during conference calls.  I'd greatly appreciate it if anyone on the 
list corrected any misunderstandings that I might have on the subject.

As an aside, how is the timing on your conference server.  The MeetMe 
application relies on it to mix the audio in conferences.  You should 
get at least 99.98% output from zttest (as shown below) or the audio 
quality will suffer.  This is an overall quality issue and is not 
necessarily related to your echo problems.

  [root at astconf ~]# zttest
  Opened pseudo zap interface, measuring accuracy...
  99.999413% 99.995407% 99.995499% 99.998047% 99.996483% 99.997849% 
99.999008%
  ...
  --- Results after 107 passes ---
  Best: 100.000 -- Worst: 99.995 -- Average: 99.997687, Difference: 
99.997815

Regards,

Matthew Roth
InterMedia Marketing Solutions
Software Engineer and Systems Developer