1611786 - [WebRTC] Poor sound quality in calls between Windows 10 and other OS's on appr.tc, Hangouts and Tokbox websites

Status: RESOLVED DUPLICATE of bug 1610647

(Core :: WebRTC: Audio/Video, defect, P1)

Description

[Negritas Sergiu, Desktop QA]

Affected versions

• 73.0b10
• 74.0a1

Affected platforms

• Windows 10 with macOS 10.13
• Windows 10 with Ubuntu 18.04, 16.04

Steps to reproduce

1. Open Google Chrome and Firefox.
2. Navigate to https://appr.tc, Hangouts or Tokbox.
3. Create a room and start a call between the 2 machines.

Expected result

• Audio and Video are clear and without interference between the 2 callers.

Actual result

• Audio quality is poor and with interference.

Regression range

• 2020-01-27T14:40:38: DEBUG : Found commit message:
  Bug 1603384 - Update cubeb from upstream to aa63601. r=padenot

Differential Revision: https://phabricator.services.mozilla.com/D56872

• Potentially regressed by: https://bugzilla.mozilla.org/show_bug.cgi?id=1603384

Comment 1

[Nico Grunbaum [:ng, @chew:mozilla.org]]

padenot, could you take a first pass at this?

Comment 2

[Paul Adenot (:padenot)]

alex, this is what you looked at the other day. Have you landed it in Gecko yet?

Comment 3

[Daniel Bodea [:danibodea]]

This issue can also be reproduced or verified using the following test page:
https://mozilla.github.io/webrtc-landing/gum_test.html
without the need of a callee.

Comment 4

[Alex Chronopoulos [:achronop]]

Yeah, this has landed with Bug 1610527 6 days ago. Can you please retest with the latest Nightly?

Comment 5

[Daniel Bodea [:danibodea]]

It still does occur on the latest Nightly v74.0a1 from 2020-01-28.

Comment 6

[Alex Chronopoulos [:achronop]]

(In reply to Bodea Daniel [:danibodea] from comment #5)

It still does occur on the latest Nightly v74.0a1 from 2020-01-28.

Thank you for checking, I am adding a NI to myself to take a look.

Comment 7

[Alex Chronopoulos [:achronop]]

To anyone reproduce this, I am not able to reproduce it. If you are in Berlin can I catch you when you have a chance to check this together? Otherwise can you please upload here your about:support page?

Comment 8

[Daniel Bodea [:danibodea]]

I have retried this issue on the latest Nightly v74.0a1 and it seems it does not occur using the steps in comment 3. Furthermore, I have attempted to use mozregression to find the fix. These are my results:
2020-01-30T17:02:42: INFO : Narrowed inbound regression window from [13c051ad, 691bd804] (3 builds) to [13c051ad, 850bc97c] (2 builds) (~1 steps left)
2020-01-30T17:02:42: DEBUG : Starting merge handling...
2020-01-30T17:02:42: DEBUG : Using url: https://hg.mozilla.org/integration/autoland/json-pushes?changeset=850bc97c56ab8542c727de1a3b361571bdfc72c8&full=1
2020-01-30T17:02:43: DEBUG : Found commit message:
Bug 1610527 - Update cubeb from upstream to 6cdf2fd. r=padenot
Differential Revision: https://phabricator.services.mozilla.com/D60518
2020-01-30T17:02:43: DEBUG : Did not find a branch, checking all integration branches
2020-01-30T17:02:43: INFO : The bisection is done.
2020-01-30T17:02:43: INFO : Stopped

Hope this helps.

Comment 9

[Julien Cristau [:jcristau]]

Do we expect bug 1610647 to fix this?

Comment 10

[Matthew Gregan [:kinetik]]

(In reply to Julien Cristau [:jcristau] from comment #9)

Do we expect bug 1610647 to fix this?

Yes, it's the same fix for beta. In Nightly, this was fixed by bug 1610527 as Daniel confirmed via mozregression in comment 8. Given this is already fixed in nightly, and now fixed in beta by bug 1610647, I'll dupe this to that bug.