Hi
I have tested echo cancellation using some different examples.
TEST 1. First, I record my voice, named TestVoice1. Then I perform testecho
using TestVoice1 as the data of both mic and speaker. Final, I get a ideal
result after modified the parameters for several times. It means the output
is almost silent. I think this shows the echo cancellation can work.
TEST 2. Then I edit the TestVoice1 using Audacity, add some other voice on
it and generate TestVoice2. This time I use TestVoice2 as the data of mic,
TestVoice1 as the data of speaker. The result is also very good.
TEST 3. First, play TestVoice1, record it as TestVoice3. Then, I adjust the
timeline of TestVoice3. So the position of sound from these two files are
almost the same. This time I use TestVoice3 as the data of mic, TestVoice1
as the data of speaker. But the result is not good. After I compared the
wave of TestVoice3 and the output found that them are almost the same, at
least very similar.
I wander if the non-linear distortion from the mic and speaker makes that
result. Then I check spectrum of TestVoice1 and Testvoice3, and there isn't
high distortion, I think. Can anyone give some advice? Thanks a lot!
BTW, I use portaudio as audio driver.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.xiph.org/pipermail/speex-dev/attachments/20060712/f2b42da7/attachment.html
