Audio and video jitterBufferDelay difference causes audio and video to be out of sync

[Qiu Chun]

Hello everyone,

I use chrome to pull from the SFU server, there may be some chrome audio and video synchronization, some chrome is not synchronized, which makes me very confused.

I don't think the reason for the out-of-sync is in the push or SFU. If they have problems then all the chromes should be out of sync.

After observation, I found that in the chrome where the audio and video are not synchronized, the audio and video jitterBufferDelay in webrtc-internals is very different.(video jitterBufferDelay/jitterEmittedCount_in_ms is around 1800 ms, audio jitterBufferDelay/jitterEmittedCount_in_ms is around 200ms)

(video)

I want to know if audio and video out of sync is caused by the difference in jitterBufferDelay? What caused the difference in jitterBufferDelay?

chrome: 77.0.3865.120(official version) 64

[Qiu Chun]

Supplementary audio and video comparison chart.

auido

video

At the same time, I found that it seems that the chrome is not synchronized when the chrome starts to flow. The video's jitterBufferDelay is very high at the beginning, and cannot be recovered during the process of pulling the stream, and the synchronized chrome will not become out of sync during the process of pulling the stream.

在 2019年10月16日星期三 UTC+8下午2:38:02，Qiu Chun写道：

[Steve Seguin]

I've been having the same issue for a few months. Problem still exists on Canary v81.  The problem does not exist on Firefox for me.

The amount of delay (jitter buffer delay) between the audio and video is sometimes 500ms to 1000ms, and then on some computers it is not present at all.

The video lags behind the audio.

The video codec used is h264.and opus audio
I am using Janus for the SFU server.

Many computers using Chrome do not have the issue, but some do.

It occurs even on very high quality Fiber optic internet connections

It happens on fast or slow computers; an Intel 9900K /w Titan RTX still faces the issue.

The video and audio are on sync for a second or so, but quickly ramps out of sync.

The issue seems to occur within our more bloated production web app, versus a stripped down version using the same computers and SFU servers.
As a result, I cannot seem to replicate the issue within a small test script, but it is definitely present in the production version of our app for many users.

It does not happen on all computers, but it can usually be reproduced on those computers that it does happen on.

If anyone has any ideas, I'd be happy to give them a try.

-steve

[Silvia Pfeiffer]

I've seen audio and video out of sync many times, but haven't had time to investigate. Thanks for sharing the analysis! It's a big problem.

Cheers,

Silvia.

--

---
You received this message because you are subscribed to the Google Groups "discuss-webrtc" group.
To unsubscribe from this group and stop receiving emails from it, send an email to discuss-webrt...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/67405615-7436-4c90-a913-40a096ba43fd%40googlegroups.com.

[Sergio Garcia Murillo]

I have been investigating this for a different issue and IMHO the jitterBufferDelay stat is incorrectly calculated for audio: 

https://bugs.chromium.org/p/webrtc/issues/detail?id=11283&can=4&q=&colspec=ID%20Pri%20Stars%20M%20Component%20Status%20Owner%20Summary%20Modified

Best regards

Sergio

--

---
You received this message because you are subscribed to the Google Groups "discuss-webrtc" group.
To unsubscribe from this group and stop receiving emails from it, send an email to discuss-webrt...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/85a6e724-4611-489d-bc20-4424350b421a%40googlegroups.com.