AEC | Double Talk | ERLE | G.168 | LEC | NLMS

Hi,all
   I have implement an AEC algorithm (Frequency Domain NLMS) on DSP (TI c5509), but when I test
it in real room the ERLE is 16dB(with white noise), will test by compositive speech source with
line_in the ERLE is 25dB.
   But, I always see that some AEC product's ERLE can get 25dB-30dB, So I want to konw the
method they test.
   1,what's the test source. white noise or speech signal?
   2,what's the test environment of there test?
   3,commonly, with a 16bit fixed-point DSP,what's the highest performance can it get?
   thanks a million!!!

(You need to be a member of echocancel -- send a blank email to echocancel-subscribe@yahoogroups.com )