Dear Sir,

I am working on Acoustic Echo Cancellation. I have
implemented both AEC without subbanding and AEC with subbanding in C+
+. For subbanding of AEC I used Polyphase IIR filter(with 70dB
attenuation).

AEC without subbanding is working fine. But in the outputs
of AEC with subbanding there exists some slight disturbance(but echo
is cancelling for all recorded audio files that are recorded through
sound recorder and for some other audio files that I got from some
application echo is not cancelling and this disturbance is continued).
So I obsereved each function(ie., polyphase lowpass filter function,
polyphase highpass filter function and the AEC operation function)
output by printing each and every output value of functions in text
files. In that I observed that where ever the highpass filter farend
sample is negative(there by estimated echo is negative and hence and
updated weights of that sample are negative) this distortion is
occuring.

So how to handle when updated weights are negative so that
to get clear echo cancelled outputs.

If any one knows information regarding this please reply to
this mail.

Thanks in advance.

Regards,
Aparna Ram.K.

"Aparna Ram" wrote in message
oups.com...
Dear Sir,

I am working on Acoustic Echo Cancellation. I have
implemented both AEC without subbanding and AEC with subbanding in C+
+. For subbanding of AEC I used Polyphase IIR filter(with 70dB
attenuation).

AEC without subbanding is working fine. But in the outputs
of AEC with subbanding there exists some slight disturbance(but echo
is cancelling for all recorded audio files that are recorded through
sound recorder and for some other audio files that I got from some
application echo is not cancelling and this disturbance is continued).
So I obsereved each function(ie., polyphase lowpass filter function,
polyphase highpass filter function and the AEC operation function)
output by printing each and every output value of functions in text
files. In that I observed that where ever the highpass filter farend
sample is negative(there by estimated echo is negative and hence and
updated weights of that sample are negative) this distortion is
occuring.

So how to handle when updated weights are negative so that
to get clear echo cancelled outputs.

If any one knows information regarding this please reply to
this mail.

Thanks in advance.

Regards,
Aparna Ram.K.


Yes, I believe we can help. Post your source code and
we can get started.

--
rb

In article .com,
Aparna Ram wrote:

Dear Sir,

I am working on Acoustic Echo Cancellation.

Snip

How exciting.

--
Telamon
Ventura, California

In article ,
"Ron Baker, Pluralitas!" wrote:

"Aparna Ram" wrote in message
oups.com...
Dear Sir,

I am working on Acoustic Echo Cancellation.

Snip


Yes, I believe we can help. Post your source code and
we can get started.

Have him email you then.

--
Telamon
Ventura, California