[mediacapture-main] Specification to capture unmodified audio

klausj has just created a new issue for https://github.com/w3c/mediacapture-main:

== Specification to capture unmodified audio ==
Current versions of Firefox (53.0.3) and Chrome (58.0.3029.110) preprocess audio capture streams for WebRTC even if the echoCancellation constraint property is set to false. I do not know what the preprocessing exactly does, but it adds some artifacts to the stream. And if we capture stereo the input streams are mixed together to mono,  preprocessed and the processed stream is duplicated. The recorded channels are always identical.  I could not find a way to capture unmodified audio using Firefox or Chrome, but this is required for some kind of applications. In our case we record speech for phonetical analysis. With this audio artifacts the recordings are not appropriate for our needs. The preprocessing is also problematic for music recording apps.
Surprisingly Microsoft Edge is the only browser I've tested, which does no audio preprocessing.

I think that Media Streams API should specify that implementations should not preprocess audio if echoCancellation is set to false.

Please view or discuss this issue at https://github.com/w3c/mediacapture-main/issues/457 using your GitHub account

Received on Wednesday, 24 May 2017 12:08:32 UTC