Loopback capture in echo cancellation

[isi...@gmail.com]

Hi,

I was looking through the WebRTC audio stack recently, and it seems that the input for the echo canceler for Windows based render/capture is the actual audio rendered

I was wondering if anyone has considered using Loopback Recording mechanism available in WASAPI, instead of using the actual audio rendered? While this would add additional complexity, this would have the advantage of removing any background noise as well from the recorded audio - this could improve audio quality considerably if audio is being rendered by the device during chat

[yoram....@gmail.com]

Yes.

We are using the WASAPI loopback to feed the speaker side of the webrtc AEC.

But we are using the webrtc AEC as standalone (with other parts of the audioprocessing package), not the whole webrtc package as is.

You can find good documentation on MSDN on the web how to read the WASAPI loopback.

Regards,

  Yoram

בתאריך יום שני, 10 באפריל 2017 בשעה 11:03:17 UTC+3, מאת isi...@gmail.com: