[pjsip] Problem with Echo Canceller on PDAs (windows mobile 5.0 / Pocket PCs)

Nanang Izzuddin nanang at pjsip.org
Wed May 21 07:08:12 EDT 2008


Hi Ramzi,

Actually for low processor client (small devices), pjmedia uses echo
suppressor instead of real echo canceler (to make sure, check whether
macro PJMEDIA_HAS_SPEEX_AEC is set to 0, for WINCE by default it is
set to 0), since real echo canceler consumes relatively high CPU load.
And echo suppressor works this way (similar to what you describes):
when speaker is playing something, the mic is turned off/muted, and
vice versa. Here is some more detail:
http://en.wikipedia.org/wiki/Echo_suppressor.

It seems the problem is as expected, however, there may always be some
ways for suppressing the frustration effect of echo suppression. It
will be great if you would share some ideas/suggestions on how to make
this echo suppressor more 'friendly'.

Regards,
nanang


2008/5/20 Ramzi Touma <rtt00 at aub.edu.lb>:
> Hi,
>
> I am having problems with the echo canceller on windows CE (PDAs / Pocket
> PCs). I tried it on several different models of these devices. I am using
> the wince_demos project, pcmu with a ptime of 40 and noVAD. I also included
> the the config_site_sample that disables the WSOLA (in case there is a
> processing issue on such small devices). I am getting unacceptable sound
> performance whenever I enable the echo canceller by setting the ec_tail to
> any value other than 0. If I were to describe what sort of sound problem I
> am encountering then it is as follows: when echo canceller is disabled, I
> get  a good sense of converstation, i.e. both ends here each other well with
> a relatively good sound quality having very few or no drop outs at all. But
> ofcourse the echo is heard whenever headphones are not used on the PDAs. Now
> when I enable the echo canceller, I notice a lot of drop outs in the
> conversation specially when a person is initially talking and the other
> tries to talk as well (probably here the echo canceller is blocking the
> microphone which has real voice and considers it as noise or something). To
> get the other person's voice to be heard, both parties should go quiet for 1
> or 2 seconds and then the first to start talking will be heard while the
> other will be blocked out (not entirely but enough to make the conversation
> unclear) .
>
>
>
> To make things clearer, if A and B are in a conversation, when A  starts to
> speak, B hears A but if B attempts to speak A while A is still talking, A
> will not hear B talking for some while; causing a lot of drop out and
> yielding the conversation useless.
>
>
>
> I hope I made myself clear somehow J
>
>
>
> Thanks,
>
> Ramzi
>
>
>
> Ramzi T. Touma
>
> Electrical and Computer Engineering
>
> American University of Beirut
>
> P.O. Box 11-0236/ 5555
>
> Riad el Solh Square 1107 - 2020
>
> Email: rtt00 at aub.edu.lb
>
>
>
> _______________________________________________
> Visit our blog: http://blog.pjsip.org
>
> pjsip mailing list
> pjsip at lists.pjsip.org
> http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org
>
>




More information about the pjsip mailing list