Residual echo activity detection.

Hi All,

Can any one help me to get documents related to Residual echo activity
detection i.e, to detect weather the output of acoustic echo cancellation
contain residual echo or not. Thanks in advance

Regards,
Neelufar.

neelufar.2005 wrote:

> Hi All,
> 
> Can any one help me to get documents related to Residual echo activity
> detection i.e, to detect weather the output of acoustic echo cancellation
> contain residual echo or not. Thanks in advance
> 
> Regards,
> Neelufar.

The cancellation is never perfect, there is always some residual error (for
the algorithms I am aware of).

As the error gets smaller, the LMS algorithm makes smaller adjustments to
the FIR coefficients until the adjustment increment is less than 1 LS bit.

In practice, the acoustic environment is always subtly changing, so the
algorithm is always hunting to try to resolve it.

neelufar.2005 wrote:
> Hi All,
> 
> Can any one help me to get documents related to Residual echo activity
> detection i.e, to detect weather the output of acoustic echo cancellation
> contain residual echo or not. Thanks in advance

You have been told many times already that there is always some amount 
of the residual echo. The question is not if there is an echo, the 
question is how much of echo is there. This can be only estimated to 
some accuracy by correlating the transmit signal and the residual signal.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com