<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<STYLE type=text/css> <!--@import url(E:\LMQ\LightingMail\\data\scrollbar.css); --></STYLE>
<META content="text/html; charset=utf-8" http-equiv=Content-Type>
<STYLE>BLOCKQUOTE{margin-Top: 0px; margin-Bottom: 0px; margin-Left: 2em}; </STYLE>
<META name=GENERATOR content="MSHTML 8.00.6001.19019"><BASE
target=_blank></HEAD>
<BODY
style="BORDER-RIGHT-WIDTH: 0px; MARGIN: 12px; BORDER-TOP-WIDTH: 0px; BORDER-BOTTOM-WIDTH: 0px; BORDER-LEFT-WIDTH: 0px"
marginheight="0" marginwidth="0"><STATIONERY>
<DIV>
<DIV>>> There is also a IEEE paper, Adaptive Sampling Rate Correction for </DIV>
<DIV>>> Acoustic Echo Control in Voice-Over-IP, which introduced a complex </DIV>
<DIV>>> method to estimate the frequency offset and resynchronize the signals </DIV>
<DIV>>> using arbitrary sampling rate conversion. I wonder if it can provide </DIV>
<DIV>>> enough performance. Because I have also designed a sampling rate </DIV>
<DIV>>> converter. After tested the offset accurately, it can reduce the </DIV>
<DIV>>> offset to less than 0.1Hz, then the signal after resampling is send to </DIV>
<DIV>>> speex AEC. But there is still hearable echo even if it is far less </DIV>
<DIV>>> than that can be heared before resampling. </DIV>
<DIV>>> </DIV>
<DIV>>> Does anybody have any suggestion about practical acoustic echo </DIV>
<DIV>>> cancellation in low-cost soundcards? You know, most low-cost </DIV>
<DIV>>> soundcards have the problem of sampling rate asynchronous. </DIV>
<DIV>>> </DIV>
<DIV>> That one sounds much more interesting. If you don't have access to IEEE </DIV>
<DIV>> papers, you can find it at the author's university site. They don't do </DIV>
<DIV>> anything extraordinary, but they have thought through how you can track </DIV>
<DIV>> the sample rate offset by LMS, and use a continuously variable rate </DIV>
<DIV>> converter to allow for it. </DIV>
<DIV> </DIV>
<DIV>Yes. As you said, this is still not a good solution.</DIV>
<DIV> </DIV>
<DIV>>
I noticed that the Fraunhofer Institute is now selling a package to </DIV>
<DIV>>
address echo cancellation when the tx sample rate cannot be trusted to </DIV>
<DIV>>
exactly match the rx sample rate - primarily in VoIP conferencing </DIV>
<DIV>>
applications. They say they use the spectral envelope, and disregard the </DIV>
<DIV>>
phase. That sounds like its not a million miles from the spectral </DIV>
<DIV>>
subtraction a lot of noise suppression schemes use, and those aren't </DIV>
<DIV>>
great at getting high levels of suppression. However, they claim very </DIV>
<DIV>>
high levels of echo suppression. There must be more to what they do than </DIV>
<DIV>> the blurb indicates. </DIV>
<DIV> </DIV>
<DIV>>
It's relatively easy to getting high levels of echo suppression using </DIV>
<DIV>>
spectral subtraction methods. In fact even half-duplex cheap </DIV>
<DIV>>
hands-free phones can achieve that. The tricky part is not to distort </DIV>
<DIV>>
the "local" voice during double-talk. That's the hard part when you </DIV>
<DIV>> can't rely on an adaptive filter. </DIV>
<DIV> </DIV>
<DIV>
<DIV>Does anybody know any kind of echo cancellation kernel which is not
sensitive to different sampling rates?</DIV>
<DIV>At least I don't know.</DIV>
<DIV>But there is still a vivid example, AEC in MSN Messager, which is a
real AEC, not a echo suppression.</DIV>
<DIV>It provides almost perfact echo cancellation even in double talk.</DIV>
<DIV>Why?</DIV>
<DIV> </DIV></DIV></DIV></STATIONERY></BODY></HTML>