[Asterisk-Users] Reducing echo on FXS port (SOLVED - Big THANKS to James Harper)

[Aryanto Rachmad]

Hello All,

The solution which is good enough for me at the moment, is James Harper's echo preload patch. The echo is reduced to the minimum from the start of the call. I just need to reduced the rxgain to make it more unnoticeable.

I implemented on SVN-branch-1.2-r8445 by hand yesterday, but I got no audio this morning. After updating to SVN-branch-1.2-r8666, the audio came back, but also the echo. So I decided to use Asterisk 1.2.1 and put in the patch.

Cheers,

Anto
  ----- Original Message ----- 
  From: Aryanto Rachmad 
  To: Asterisk Users Mailing List - Non-Commercial Discussion 
  Sent: Monday, January 23, 2006 12:13 AM
  Subject: Re: [Asterisk-Users] Reducing echo on FXS port


  Hello Giovanni and everybody,

  Thanks a lot for your suggestion.

  Unfortunately, that does not help. With READ_SIZE=16, I got vibrating voice on the speaker phone. With READ_SIZE=80, the voice came back to normal and the echo is more reduced but still noticeable. I finally changed it back to 160 as it seems to affect a lot of things in chan_zap.

  I am sorry that I didn't properly explain the setup I have. I think this issue happens because I use wireless handset. Here is my actual setup:

                                                           ---------------  out ------------
                                                           |             | <--- |          |
  Wireless handset <--> Base phone <--> FXS (TDM400P) <--> | ZAP Channel |      | Asterisk |
    (Mic muted)                                            |             | ---> |          |
                                                           ---------------  in  ------------
                                                                             ^
                                                                          Monitor()

  The delay between the original voice going out from asterisk to the phone, and its echo coming back to asterisk is about 104 ms, assuming that Monitor() application wrote both files at exactly the same time. How did I find that? I loaded the "in" and "out" files created by Monitor() application into Audacity (audacity.sourceforge.net). I could not find any other method to find this delay. Does anyone know a better and more accurate method?

  When I did the same thing using X-Lite with below setup:

                   ---------------  out ------------
                   |             | <--- |          |
  PC (X-Lite) <--> | SIP Channel |      | Asterisk |
  (Mic muted)      |             | ---> |          |
                   ---------------  in  ------------
                                     ^
                                  Monitor()

  There is no sound at all on the "in" file created by Monitor() application, indicating that there is no echo at all as the microphone on X-Lite was muted.

  Since the echo on FXS is consistent, there must be a way to eliminate it. The question is how? I think it can not be done only by changing the configuration parameters related to echo, especially with this huge delay.

  Do you have any other suggestions?

  Cheers,

  Anto
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.digium.com/pipermail/asterisk-users/attachments/20060125/d3c2e2bb/attachment.htm