Thanks, Steve, Rocky and MMHafezi for excellent explanations.  From my own 
experience, I knew that echo cancelling was more critical (and the echo more 
annoying) with longer delays, but didn't put all the pieces together, especially 
the fact that VOIP may involve 2-wire POTS.

To the OP, I can't help with C source other than to suggest a Google search as I 
think I've seen similar questions pop up here before.

-- 
Jon Harris
SPAM blocker in place:
Remove 99 (but leave 7) to reply


"Steve Underwood" <steveu@dis.org> wrote in message 
news:dm227d$88m$1@nnews.pacific.net.hk...
> Jon Harris wrote:
>> Does VOIP have some innate need for echo cancellation?  Or is the echo that 
>> needs to be cancelled just the acoustic echo of a speakerphone-type system 
>> that happens with any communications link?  I would think the latter, since 
>> with VOIP the transmit and receive signals are electrically independent no 
>> (as opposed to one a phone where there both sides go down the same twisted 
>> pair)?
>>
> VoIP has a very serious need for echo cancellation. Because of the latency in 
> VoIP connections, any echo from a hybrid on a PSTN interconnect - even 
> earpiece to mic coupling in a handset - becomes a seriously annoying echo. In 
> normal PSTN work these things merely give a little pleasant reverberation. 
> VoIP is just the same in this regard as digital cellular systems, which also 
> have a very strong need to eliminate echoes before signals pass across the 
> high latency section of the path.
>
> Steve

Jon Harris wrote:
> Does VOIP have some innate need for echo cancellation?  Or is the echo that 
> needs to be cancelled just the acoustic echo of a speakerphone-type system that 
> happens with any communications link?  I would think the latter, since with VOIP 
> the transmit and receive signals are electrically independent no (as opposed to 
> one a phone where there both sides go down the same twisted pair)?
> 
VoIP has a very serious need for echo cancellation. Because of the 
latency in VoIP connections, any echo from a hybrid on a PSTN 
interconnect - even earpiece to mic coupling in a handset - becomes a 
seriously annoying echo. In normal PSTN work these things merely give a 
little pleasant reverberation. VoIP is just the same in this regard as 
digital cellular systems, which also have a very strong need to 
eliminate echoes before signals pass across the high latency section of 
the path.

Steve

Have a look at Asterisk.

John

Hi Harris,

Thanks for your reply. No, the echo is not from speakerphone. In fact, one
side, is over ip while the other side is on a common two wire phone line.
Actually, the voice from the talker on computer side passes a local area
network and feeds a hardware that converts the signal to analog and puts
it on a two wire phone line, here a return signal feeds back to the D/A
and through the network comes back to talker and because of the long delay
it is very disturbing.

Now, I want to develope an Echo Canceller software on computer side to
cancel this echo. I should emphasize that on computer side I just have
access to sent and received buffers of voice samples.

I really appreciate any hint from you.

regards,

M. Hafezi

>Does VOIP have some innate need for echo cancellation?  Or is the echo
that 
>needs to be cancelled just the acoustic echo of a speakerphone-type
system that 
>happens with any communications link?  I would think the latter, since
with VOIP 
>the transmit and receive signals are electrically independent no (as
opposed to 
>one a phone where there both sides go down the same twisted pair)?
>
>-- 
>Jon Harris
>SPAM blocker in place:
>Remove 99 (but leave 7) to reply
>
>"MMHafezi" <MMHafezi@Gmail.com> wrote in message 
>news:f8CdnfSVrt4cXhzenZ2dnUVZ_v2dnZ2d@giganews.com...
>>
>> Hi everybody,
>>
>> Does anybody know any C source code for Echo Cancellation on a voice
over
>> ip system?
>>
>> Regards,
>>
>> M. Hafezi 
>
>
>

Hi Jon,

Jon Harris wrote:
> Does VOIP have some innate need for echo cancellation?  Or is the echo that
> needs to be cancelled just the acoustic echo of a speakerphone-type system that
> happens with any communications link?  I would think the latter, since with VOIP
> the transmit and receive signals are electrically independent no (as opposed to
> one a phone where there both sides go down the same twisted pair)?

Often the 'last mile' of a VIOP link is to a 2 wire phone. The echo is
normally caused by the less-than-infinite transhybrid loss.

Regards
Robert

Does VOIP have some innate need for echo cancellation?  Or is the echo that 
needs to be cancelled just the acoustic echo of a speakerphone-type system that 
happens with any communications link?  I would think the latter, since with VOIP 
the transmit and receive signals are electrically independent no (as opposed to 
one a phone where there both sides go down the same twisted pair)?

-- 
Jon Harris
SPAM blocker in place:
Remove 99 (but leave 7) to reply

"MMHafezi" <MMHafezi@Gmail.com> wrote in message 
news:f8CdnfSVrt4cXhzenZ2dnUVZ_v2dnZ2d@giganews.com...
>
> Hi everybody,
>
> Does anybody know any C source code for Echo Cancellation on a voice over
> ip system?
>
> Regards,
>
> M. Hafezi