Audio distorted after 30 seconds

[Dylan-Buth]

We've been using this package for a while and just recently ran into an issue where after 30 seconds, the audio is distorted. The first 30 seconds sound perfectly fine. Is this something you've heard of before? Or have any advice on debugging it?

[chris-rudmin]

I have not encountered this. For debugging, I would recommend to see if you can narrow it down to a specific platform / browser. If it occurs consistently, I would add some breakpoints and check your audio data, see if the browser is sending distorted audio data.

[chris-rudmin]

Enable monitoring (with headphones on). If your monitoring contains distorted audio, then it is the browser sending distorted audio. If the monitoring signal is clean, but the encoded data is distorted, then it would suggest a bug in opus-recorder.

[rotem925]

This is a bug in the recent release of Chrome and does not happen on Canary, see https://bugs.chromium.org/p/chromium/issues/detail?id=1381319&q=30%20seconds&can=2
I guess that's the issue... it's not happening only with opus-recorder.

[z-briand]

We're seeing this issue as well; it's been a low-level issue for a while, but coincidentally I searched to see if anyone else was seeing this last week, and hit Dylan's ticket.

I had told my team we should file a ticket, and they said they already had -- Dylan's bug is actually this guy:
#176
So the interesting thing there is that this issue has been lurking for years.

I had my team create a standalone test environment, so that we could rule out other aspects of our platform. We had 12 people test; 3 could replicate pretty consistently (it doesn't happen every clip, but if you record a handful, you'd definitely get at least one with glitches), and 9 that couldn't replicate it ever. So there's definitely something about the machine config that matters here.

Our hypothesis is that it happens when a machine is short on RAM, but we're still trying to get a smoking gun on that.

[z-briand]

OK, all of our testers who replicated this were on Chrome. Many of our Chrome users couldn't replicate the bug, so it's probably actually Chrome + RAM shortage, or something similar? None of our non-Chrome users could replicate.

All of that is consistent with Rotem's pointing the finger at Chrome, and the timing of Dylan's report suggests he's hitting the Chrome bug, not verifying that #176 is unresolved. It's possible #176 is still out there, but with the Chrome bug in the field, all of our recent data collection on our end isn't conclusive.

At this point, I think you guys can close this ticket? If my team sees any issues after the Chrome team releases 109.0.5366.0, then we'll do a better job at finding you a smoking gun than we did in the #176 era.

[Dylan-Buth]

For the record, what I'm experiencing just seems to be the chrome bug. Works on safari just fine, and I have plenty of RAM so I don't think it's related to that. For me anyway.

I appreciate you all responding here. I would've spent hours debugging this before finding the open Chrome issue.