Closed ipatix closed 2 years ago
I generated a sine wave as you described and then exported it as 32-float .wav (among other options). The bug cannot be reproduced, maybe it's been already fixed.
Well, I haven't been able to try it again, but my instructions theoretically might not be sufficient to reproduce this problem since depeding on the caluclation accuracy a sine wave might not contain a critical value (like in the image not all cycles have this phenomenon).
Since I usually do not build Polyphone from source, which version should I try to see if this problem still exists?
Fixed! I reproduced the bug with an amplified sample (using Audacity and allowing saturation). IEEE float wav files possibly contain values exceeding 1.0 and I didn't take it into account.
This bug, which is causing very nasty sounds, is very easy to reproduce:
This is not a playback problem but an actual sample data problem. When I export the sine wave back to Audacity, you can see the following difference:
So most likely the maxmimum positive float value is improperly converted to int and causing an overflow.