Bug (Miguel) – BiquadFilter stops working correctly after 512 seconds

• Component: Web Audio
• Status: RESOLVED INVALID

[Miguel]

While doing some bass line loop tests, I noticed that the low pass filter stopped to work as expected after 512 seconds.
I'm using Chrome version 20.0.1132.57, I hope It's ok to fill the bug here.

[Raymond Toy]

(In reply to comment #0)
> While doing some bass line loop tests, I noticed that the low pass filter stopped to work as expected after 512 seconds.
> I'm using Chrome version 20.0.1132.57, I hope It's ok to fill the bug here.

This is the right place.

What do you mean by stopped working as epxected? Do you have a small example that illustrates the problem?  That would be best.  I ran a small demo I have for 10 minutes and the low pass filter works fine.

[Miguel]

Created attachment 155607 [details]
A test describing the bug

Everytime I run this, the filter (I suppose) stops to work after 512 seconds. I hope it's not related to my code.
Included is a timer showing you the elapsed seconds taken from AudioContext.currentTime.

[Raymond Toy]

(In reply to comment #2)
> Created an attachment (id=155607) [details]
> A test describing the bug
> 
> Everytime I run this, the filter (I suppose) stops to work after 512 seconds. I hope it's not related to my code.
> Included is a timer showing you the elapsed seconds taken from AudioContext.currentTime.

It seems to be something to do with your code.  I replaced the line s.connect(l) with s.connect(a.destination) to remove the lowpass filter from the audio graph.  Audio suddenly stops after approximately 512 sec.

There could be other bugs in webaudio, but it seems that the lowpass filter is not the cause of this particular issue.

If you could simplify your code and still reproduce the issue with audio stopping after 512 sec, please let us know.

[Miguel]

Created attachment 155816 [details]
Simplified test

Here is a simplified version where the issue appears again.

[Raymond Toy]

(In reply to comment #4)
> Created an attachment (id=155816) [details]
> Simplified test
> 
> Here is a simplified version where the issue appears again.

Thanks for the new simplified test.  Since this test does not use a biquad filter, I'm going to close this bug report and open a new one with this test case.

[Miguel]

I continued my tests, and I believe the issue is related to exponentialRampToValueAtTime function. The test works when I remove it.

[Raymond Toy]

(In reply to comment #6)
> I continued my tests, and I believe the issue is related to exponentialRampToValueAtTime function. The test works when I remove it.

Thanks for looking into it!  I'm closing this bug. See https://bugs.webkit.org/show_bug.cgi?id=92875.