4
$\begingroup$

AEC algorithms mostly rely on LMS adaptive filtering, i.e. you update FIR filter coefficients then perform the filtering. Theoretically, the FIR must be as long as the maximum echo length you want to cancel. For instance to cancel delays up to 500ms on a 48kHz signal, you'll need a 24000 point FIR. When your memory and processing power limitations make it so you can neither afford to perform 24000 MACs per processed sample, nor to use FFT-based fast convolution algorithms, is there a way around for cancelling such potentially high length echoes in a more affordable way, given delay is unknown and potentially variable?

I was wondering maybe about some other algorithm running in parallel that could assess the approximate delay, then use an adaptive length delay line + a shorter adaptive FIR filter (up to a few hundreds taps is OK)

Does this make sense? Any other neat approach to suggest?

$\endgroup$
  • $\begingroup$ Good question. Do you expect there to be one dominant echo that you want to cancel, or multiple ones? I've never had to mess with echo cancellation before, but something like what you suggested sounds like it could work. You could use cross-correlation, for instance, to coarsely assess the bulk delay, then center your echo canceller about that offset. $\endgroup$ – Jason R Oct 27 '15 at 13:51
  • $\begingroup$ Let's get started with a single echo... Right, if there are several ones, the variable delay trick won't work for reducing FIR length. As for assessing the delay, I was indeed thinking about things like cross correlation $\endgroup$ – pat Oct 27 '15 at 17:29
-1
$\begingroup$

I believe you logic is reasonable. The fact that there are companies offering algorithms for long echo tail (even above 1 second) with reasonable performance indicates that an algorithm similar to your suggested logic can work.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.