1268403 - Noticeable background noise when sharing microphone

Status: RESOLVED INCOMPLETE

(Core :: WebRTC: Audio/Video, defect)

Description

[Ada [:adalucinet]]

[Affected versions]:
- latest Developer Edition 48.0a2
- latest Nightly 49.0a1

[Affected platforms]:
- Mac OS X 10.9.5
- Ubuntu 16.04 x64

[Steps to reproduce]:
1. Launch Firefox.
2. Load the Web Audio API demo page - http://goo.gl/TlJfJB
3. Click on Add Audio Source button
4. Select Live Input
5. Share the microphone when prompted and connect the node

[Expected result]: The sound is properly played back.

[Actual result]: Background noise present.

[Regression range]:
- (m-c):
Last good revision: 973dfa15822126c39ea2d98d34ac16643b040243
First bad revision: f05a1242fb29023bd7ebc492897ed3d6907733c7
Pushlog: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=973dfa15822126c39ea2d98d34ac16643b040243&tochange=f05a1242fb29023bd7ebc492897ed3d6907733c7
- (m-i):
Last good revision: 68a682ef9e59eabb5d0f1fda05e84a773ca8340b
First bad revision: 7f2f44eaea09c02079d8f1a0be777cd39776712e
Pushlog:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=68a682ef9e59eabb5d0f1fda05e84a773ca8340b&tochange=7f2f44eaea09c02079d8f1a0be777cd39776712e

Looks like the following bug has the changes which introduced the regression: https://bugzil.la/1243857

[Additional notes]:
- Reproducible with e10s on/off
- Unable to reproduce under Windows 10 64-bit and Mac OS X 10.11.1
- Also reproducible with https://mozilla.github.io/webrtc-landing/gum_test.html test page, but only with Audio; it's properly working with both Audio & Video

Comment 1

[Randell Jesup [:jesup] (needinfo me)]

Thanks - can you tell us what audio input and output devices are in use?  Note that this test will cause the echo canceller to try to cancel your voice out, partially, resulting in some odd artifacts.

It would help a lot to get AEC logs from before the regression and after.  In a non-e10s window: start the WebAUdio test.  Open a second tab, and go to about:webrtc.  Select "Start AEC Logging"  Run the test for a short while (15-30 seconds), and talk some.  Then select Stop AEC Logging, and grab all the aec* files from the indicated directory, compress them in a tar/zip/etc, and add them here (or put them on dropbox/etc where I can grab them).

This should let me hear what the noise is.  I'm unable to reproduce any problem with a Logitech 540 USB headset on Fedora64 22.  I don't have a 10.9 mac.

Thanks

Comment 2

[Alex Chronopoulos [:achronop]]

I was not able to reproduce on Ubuntu 16.04 x64 on mac hardware

Comment 3

[Maire Reavy [:mreavy]]

One more question for Alexandra: Can you explain how the behavior differs from Fx47 to Fx48 -- since you tagged this as a regression?  Thanks.

Comment 4

[Alex Chronopoulos [:achronop]]

Can you please also check in case of mac if the "Use ambient noise reduction" option in on or off during your tests. This should be at System Pref -> Sound -> Input.

Comment 5

[Ada [:adalucinet]]

Unfortunately, I'm not able to reproduce this issue - tried with latest Aurora and Nightly builds, even with the build and profile it reproduced in the first time; Is it possible to be a site issue? What do you guys think?

Comment 6

[Randell Jesup [:jesup] (needinfo me)]

(In reply to Alexandra Lucinet, QA Mentor [:adalucinet] from comment #5)
> Unfortunately, I'm not able to reproduce this issue - tried with latest
> Aurora and Nightly builds, even with the build and profile it reproduced in
> the first time; Is it possible to be a site issue? What do you guys think?

It's unlikely (though not impossible) for it to be a site issue.  Are you using the exact same hardware and OS settings?  (same headset, same laptop, etc)?

Comment 7

[Ada [:adalucinet]]

(In reply to Randell Jesup [:jesup] from comment #6)
> (In reply to Alexandra Lucinet, QA Mentor [:adalucinet] from comment #5)
> > Unfortunately, I'm not able to reproduce this issue - tried with latest
> > Aurora and Nightly builds, even with the build and profile it reproduced in
> > the first time; Is it possible to be a site issue? What do you guys think?
> 
> It's unlikely (though not impossible) for it to be a site issue.  Are you
> using the exact same hardware and OS settings?  (same headset, same laptop,
> etc)?

Yes, everything is exactly the same.

Comment 8

[Randell Jesup [:jesup] (needinfo me)]

thanks.  Given this, let's close as incomplete for now.  Reopen if you find it again.