Random bug crashes audio and causes infinitely LOUD noise

Having trouble with LMMS? Ask about it here.
So I'm reporting this bug that's been quite common these days:

- LMMS 1.0.0, playing (or recording) a song containing 2 TripleOscillator presets, plus other instruments with lots of FX

- Randomly, one of those TripleOscillator presets will CRASH, and crash all LMMS audio output.

- When it crashes, I can see, with the red lights in the corresponding mixer, that the TripleOscillator channel goes into Infinitely Loud Silence, or sometimes worse, infinitely loud WHITE NOISE ! :o I almost had heart attacks or became deaf several times because of that.

- Using certain configuration on Calf Compressors, might increase the bug's occurence, but I'm not even sure. It happens without compressors as well.

- The TripleOscillator presets I always use are SEGuitar and Erazzor, so a Clip plugin is involved, and I'll most likely add the VyNil plugin above it. When I stop playing a song with a VyNil enabled anywhere, it keeps making ambient scratch noises when the song is stopped, so the infinitely loud bug also happens when stopped if there's an active VyNil on the TripleOscillator channels.

- It's totally random, not CPU-related etc., and I have never been able to reproduce that bug manually.

- This doesn't happen when exporting the song. It used to, but it doesn't anymore. It seems like something has been corrected, well done. :)

Could it be one of those LADSPA plugins, which causes a NaN value in the output as they discussed in another thread here ? Does anybody knows anything about this kind of bug, so we could prevent it ?
This issue https://github.com/LMMS/lmms/issues/1048
i think?
So a ticket exists already. Thanks for reporting.
You're welcome.

By the way, I was actually able to identify the buggy plugin in the case of infinitely LOUD noise, and it was C*Plate 2x2 Reverb. I managed to deactivate the plugin when the loud noise had started, and the music was playing back normal. Switching the plugin on again started the loud noise again.

So I will try working on my tracks without reverb, and see if it solves part of the problem.
I'm debugging right now and observing LMMS while writing these words, and I may have identified the cause of LOUD noises and ILS, and that it actually doesn't crash the whole LMMS audio.

I did further research and found that when I switch off and on either C*Plate 2x2 or Calf compressors while a loud noise is playing, it interrupts the sound of white noise. I also noticed that a loud noise always comes from one of the TripleOscillator tracks (but its repercussion is on the master track), and that it decreases with time, eventually fading out and the song playing back normally, and that it randomly pops up several times while playing a song, and also while doing nothing , as there's still sound coming from the VyNil plugins which only are on... TripleOscillator tracks, i.e. where loud noises are generated.

And if you remember what I said before about the VyNil plugin, I'm thinking of a bug implying all those 3 plugins. Like if there's a VyNil on a channel that generates random clicks and pops, and a C*Plate Reverb or compressor on the master, I can imagine that loud noises are random VyNil clicks and pops which would be amplified to +Inf dB (or "NaN dB") by reverbs and compressors on the master track. So according to my observations:

VyNil + Master C*Plate + (no compressor) = LOUD NOISES
VyNil + Master C*Plate + compressor = LOUD silences
VyNil + (no reverb) + (no compressor) = Here I'm hearing strange "square frequence" beeps popping up randomly, at random frequences. They only come from the two TripleOscillator tracks with VyNil. Those beeps could be the cause of loud noises, and as they also pop up when song is finished, they can only come from VyNil.

The last thing I'm trying is to get rid of the VyNil plugins on TripleOscillator tracks and trying again with Compressor and Reverb on. No beeps and loud noises so far. So it's solved for me :D (I hope so) unless I come back with more information.

EDIT: to sum up in three words, "Deactivated VyNil : Solved!"
EDIT: to sum up in three words, "Deactivated VyNil : Solved!"
:) all is well, i guess -But i do have one more question:

When you experience this bug, does the Wave / CPU Display Image
has this behavior:
First it has an almost horizontal line, that oscillates slowly up and down, then it turns red and disappear. Almost simultaneous, all sound disappear?

Is that the case?
In case of silence, the line disappears completely at the same moment the sound goes off. The line doesn't turn abnormally red, it just disappears. All FX channel volume meters run normally except the master "0" which has no signal at all.

In case of loud noise without silence, the line is TOTALLY RED and the wave is square. Then after the loud noise has gradually decreased, the line gets normal again (like when a song is playing) until the next "blast" happens.

Whether there's a silence or a loud noise, it depends on whether the Calf Compressor added above "VyNil + Reverb" is enabled (silence) or disabled (loud noise).
jurawolfproductions wrote:In case of silence, the line disappears completely at the same moment the sound goes off. The line doesn't turn abnormally red, it just disappears. All FX channel volume meters run normally except the master "0" which has no signal at all.
That was exactly what i expected, and i can tell you why:
If a low note has too high resonance, the loudness will make the oscillation go outside the area that is covered in the Wave / CPU Display
There is in fact still a sound, but the oscillation is so slow, that the line not only is next to horizontal, it also oscillates so slow, that it appear to be 'stuck'.
This is not trivial, because both very high, and very low sound, can be bad for your hearing, and your speakers, but hearing is worse.
You get this because you have to much resonance, and loudness. If you cut the resonance and loudness, you would be able to see a seemingly horizontal line move very slowly up and down. It is in fact just a tiny part of a very long wave, but be careful, with sounds like that!