Closed Bug 1628779 Opened 5 years ago Closed 4 years ago

Investigate changing our audio input processing settings

Categories

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

Product:
Core
Component:
WebRTC: Audio/Video
Type:
enhancement

Tracking

()

Status:
RESOLVED FIXED
Milestone:
mozilla78
Tracking Flags:
Tracking Status
firefox78 --- fixed

People

(Reporter: padenot, Assigned: padenot)

References

(Depends on 1 open bug, Blocks 1 open bug, Regressed 2 open bugs)

Details

Attachments

(4 files, 3 obsolete files)

Bug 1628779 - Add a function to get an estimation of the roundtrip latency of the default devices. r?achronop
(deleted), text/x-phabricator-request
Details
Bug 1628779 - Add a method on DOMWindowUtils to query an estimate of the round trip latency of the default audio devices. r?chunmin
(deleted), text/x-phabricator-request
Details
Bug 1628779 - Add an entry in about:support for the roundtrip audio latency of default devices r?chunmin
(deleted), text/x-phabricator-request
Details
Bug 1628779 - Update audioipc to 1b27476950.
(deleted), text/x-phabricator-request
Details

It's time to reevaluate our input processing, to see if we can get better results out of it.

In particular I'm not happy about the AGC, but I'll look at others.

Outside of bug 1627313 (that has a patch), I find the echo cancellation works well, but I'd like to give it the real round-trip latency, this can help with performances. I'm going to tweak it and try to reproduce hard cases (people moving, big rooms, etc.) that happen in real-life.

I haven't assessed noise suppression yet, but I've anecdotally had instances where it wasn't performing as good as I thought it should have. Maybe it's worth having a look at rnnoise again.

Paul can you set P2 or P3? Not sure if you mean to work on this right away or later.

Flags: needinfo?(padenot)
Flags: needinfo?(padenot)
Priority: -- → P1
Depends on: 1635404
Depends on: 1635497
Depends on: 1626081

We're doing it during stable state so script can't observe the change. They are
cached in atomic variables, so that the rendering thread can have the info as
well, for using during processing.

Assignee: nobody → padenot
Status: NEW → ASSIGNED

This is blocking, and will be used off-main-thread.

Attachment #9149051 - Attachment description: Bug 1628779 - Update the latency periodically on main thread. r?achronop → Bug 1628779 - Trigger latency updates periodically from main thread, in the background. r?achronop
Attachment #9149051 - Attachment description: Bug 1628779 - Trigger latency updates periodically from main thread, in the background. r?achronop → Bug 1628779 - Update the latency periodically on main thread. r?achronop
Pushed by padenot@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/61336bfc7584 Update the latency periodically on main thread. r=achronop https://hg.mozilla.org/integration/autoland/rev/99be85a6ec5c Add a function to get an estimation of the roundtrip latency of the default devices. r=achronop https://hg.mozilla.org/integration/autoland/rev/eb232dca19b0 Add a method on DOMWindowUtils to query an estimate of the round trip latency of the default audio devices. r=chunmin https://hg.mozilla.org/integration/autoland/rev/5e0b79c5f7e1 Add an entry in about:support for the roundtrip audio latency of default devices r=chunmin,fluent-reviewers,flod https://hg.mozilla.org/integration/autoland/rev/cc82aacadf80 Pass the audio rountrip latency to the echo canceller, and disable extended filter and delay agnostic AEC, on macOS Nightly. r=achronop https://hg.mozilla.org/integration/autoland/rev/016ee047a191 Update audioipc to 1b27476950. r=cubeb-reviewers,kinetik https://hg.mozilla.org/integration/autoland/rev/5e969a4bab42 Disable getting the input and output latency on Linux. r=achronop
Regressions: 1642712

I'm not sure yet but it seems that this patch broke the video on macOS.

https://bugzilla.mozilla.org/show_bug.cgi?id=1643411

Regressions: 1643411
Regressions: 1645755
Depends on: 1652595
Regressions: 1653113
Attachment #9151839 - Attachment is obsolete: true
Attachment #9149056 - Attachment is obsolete: true
Attachment #9149051 - Attachment is obsolete: true
Blocks: about-media
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: