[Asterisk-Users] Echo Canceller question- is there a viable solution?

[Matthew Fredrickson]

On Oct 27, 2005, at 12:38 AM, gw at adcomcorp.com wrote:
> My question is, what is the direction in relation to analog boards and
> such?

Right now, it looks like the current fad of the asterisk group is 
hardware echo
cancelation.  However, there is work that is occurring on the software 
echo
cans to improve them.  In fact, I just committed basically an update to 
KB1
(which was until now the latest and greatest version of MEC2) that is 
supposed
to provide somewhat significant improvements.

>
> Quite a few people tend to have difficulties with echo, and although 
> the
> WIKI has some very helpful advice, from a business standpoint I would
> think that it would be an important step to come up with a final
> solution to the problem.
>
> Many companies who make the higher end equipment seem to have tackled
> the issue on their hardware.
>
> Do we know if digium is spending time on solving the issue?  For
> example, having a tool to run on a digium analog or t1 board to analyze
> the line statistics and come up with the proper gain settings could be
> extremely helpful.
>
> Such a tool would require a firm knowledge of the causes and solutions
> to echo however, but I would assume that digium should have a grasp on
> this.
>
> It just seems difficult to suggest to companies to use an asterisk 
> based
> solution (if they do not use pri) when there is the possibility that an
> installation will have issues with echo.
>
> At this point, it feels more like a trial experience to eliminate echo
> in various environments.

Unfortunately, that's the way it is right now.  Getting to the point 
where you have
enough knowledge to be able to work on these things is not an 
insignificant task.
It seems like we're slowly getting there, and now that we have some 
more interest
on improving the software echo cans we might be a little be closer to 
getting to the
point where it "just works".

>
> I have used local tone from the CO to help narrow things down, but a
> tool that would loop dial a line and do an analysis could reduce the
> implementation time from days to hours.

Well, there isn't anything that does the "whole job" right now.  
There's a bunch
of pieces that go together, and if you have the necessary knowledge of 
how to
put the pieces together, you can get pretty close to it "just working". 
  It's not that
bad though, one can also see it as job security as well :-)

Matthew Fredrickson