New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 845880 link

Starred by 3 users

Issue metadata

Status: Untriaged
Owner: ----
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 2
Type: Bug-Regression



Sign in to add a comment

getUserMedia() with explicit deviceId yields silent audio track with BT headsets on CrOS every 2nd time

Reported by thembr...@gmail.com, May 23 2018

Issue description

UserAgent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_13_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/66.0.3359.181 Safari/537.36
Platform: 10682.0.0 (Official Build) dev-channel kevin

Steps to reproduce the problem:
1. Open https://output.jsbin.com/hobuhi on CrOS with a BT headset connected
2. If sources is empty, click 'get stream' version, grant access and re-load page.
3. Select bluetooth headset in dropdown explicitly (NOT via default input)
4. Click on 'get stream' and watch console log for 'hasAudio' messages

What is the expected behavior?
Audio streams should contain audio data consistently.

What went wrong?
Audio stream is silent / does not contain any data on exactly every second attempt it is acquired. It does not matter if the page is re-loaded in-between.

Did this work before? Yes 64 (did not verify)

Does this work in other browsers? Yes

Chrome version: 68.0.3431.0  Channel: dev
OS Version: 
Flash Version: 

The demo uses the web audio API, however this not a WebAudio issue, feeding the stream into MediaRecorder instead of the WebAudioAPI shows the same symptoms (recorder never yields any data when stream is 'silent').

Model of BT headset does not seem to matter, reproduces on all I tried (e.g. Airpods).

Source is here: https://jsbin.com/hobuhi/edit?html,js
 

Comment 1 by guidou@chromium.org, May 24 2018

Components: -Blink>WebRTC Blink>WebRTC>Audio Internals>Media>Capture
Cc: phoglund@chromium.org
Owner: guidou@chromium.org
Status: Assigned (was: Unconfirmed)
Patrik, can someone from your team help us reproduce and confirm this bug, please?
Cc: vasanthakumar@chromium.org rantonysamy@chromium.org
Wow, that is an exotic bug. Ravi, Vasan, can you shoehorn this one in during your CrOS test passes? Marina, what logs do you want if they repro?
Sure we will get reproduced with our test environment. 
Thank you. Audio traces would be useful, from chrome://tracing, and PeerConnection updates and stats data, from  chrome://webrtc-internals. Guido, anything else that might be useful?
For now, that should be enough. The most important thing would be to be able to reproduce it.
Hi all, we have reproduced the issue on M68 68.0.3440.4. 

Kindly find the logs in this link: 
https://cloud.google.com/console/storage/chromiumos-test-logs/bugfiles/cr/845880/

We will reproduce this issue in M64 or older version to confirm if this is a regression. We will also add this test in our smoke test suite.  
This issue is reproducible in M64 64.0.3283.190 in Kevin. Thus, It can be a old bug.

This issue is reproducible in Santa device as well. It is not a device specific issue.
Can you try on another platform just to confirm that it's a platform-specific bug?
@Guid: I have tried to reproduce this issue in Macbook Pro 15" and windows 10 Asus  UX32A. It works as intended. 

I don't have bluetooth dongle to check in Linux work station. I will update here as soon I get one. 

Thus a crOS specific as of now. 
Cc: guidou@chromium.org
Components: -Internals>Media>Capture -Blink>WebRTC>Audio OS>Kernel>Audio
Owner: ----
Status: Untriaged (was: Assigned)
Switching component to OS>Kernel>Audio since this is CrOS specific.
Please switch the component back if this turns out to be incorrect.

Sign in to add a comment