speex-devDear Jean-Marc Valin
I got some problems with evaluating the AEC module of speex. I wrote a test
main function and compiled it with the speex lib in VC6.0, it initialized the
AEC state and called the AEC main function in the same way as what was done in
testecho.c. The near-end input wave file was a simple delaying and adding
version of the far-end input wave, eg. y(n) = 0.5*x(n-100)+ 0.2*x(n-200)+
0.1*x(n-300) . The max delay was ensured to be less than the setting echo tail
length of AEC, but the output e(n) was not an echo canceled verson of x(n), It
seemed that the ^h(n) never converged to the real impulse response h(n), why?
and I didn't suspect the input files because the same x(n) and y(n) input
could work well in another simple time-demon nlms AEC method which was
implemented in malab.
????????
best regards,
???????? Johnson Peng
????????p_yuanjiang@sina.com
??????????2006-09-29
The AEC is not designed to cancel sinusoids. Try with speech. Jean-Marc Johnson Peng a ?crit :> speex-devDear Jean-Marc Valin I got some problems with evaluating the > AEC module of speex. I wrote a test main function and compiled it > with the speex lib in VC6.0, it initialized the AEC state and called > the AEC main function in the same way as what was done in testecho.c. > The near-end input wave file was a simple delaying and adding version > of the far-end input wave, eg. y(n) = 0.5*x(n-100)+ 0.2*x(n-200)+ > 0.1*x(n-300) . The max delay was ensured to be less than the setting > echo tail length of AEC, but the output e(n) was not an echo canceled > verson of x(n), It seemed that the ^h(n) never converged to the real > impulse response h(n), why? and I didn't suspect the input files > because the same x(n) and y(n) input could work well in another > simple time-demon nlms AEC method which was implemented in malab. > > > ???????? > > best regards, > > ???????? Johnson Peng ????????p_yuanjiang@sina.com ??????????2006- > 09-29 > > > ------------------------------------------------------------------------ > > > _______________________________________________ Speex-dev mailing > list Speex-dev@xiph.org > http://lists.xiph.org/mailman/listinfo/speex-dev