search

LMMS / lmms

Cross-platform music production software
https://lmms.io
GNU General Public License v2.0
8.15k stars 1.01k forks source link

Instruments with envelope/LFO amount close to 1.0 under cutoff tab causes sound to crash #468

Closed karmux closed 10 years ago

karmux commented 10 years ago

LMMS loses sound when instrument (TripleOscillator, BitInvader, Organic, ...) has envelope or LFO amount close to 1.0 under cutoff tab. To reproduce this glitch a filter frequency must be relatively high too. It sounds like some kind of overflow. To get sound back I have to reopen the project. Added screenshot of TripleOscillator after sound crash. I'm using LMMS 0.9.92. tosc

musikBear commented 10 years ago

perhaps you could save the preset and upload it ? - i think that what you see is serious clipping (the visualisation-window shows red oscilations? right?) and, yes then lmms can 'jam' but try to disable the synth - does the sound not come back then? If it indeed what i think it is, then it should be worse in low register, and difficult to induce in high register -yes/ no? -and uh-oh! btw be carefull with loudness - you can damage both your speakers and worse -your ears!

zonkmachine commented 10 years ago

To get sound back I have to reopen the project.

I'm seeing the same but I don't have to reopen the project after the glitch. I'll try and submit an example later today.

zonkmachine commented 10 years ago

Also, I can work around it by exporting with higher quality settings.

Edit! Which makes me think this could have something to do with pushing the frequency up to 'half the sample rate'

karmux commented 10 years ago

As musikBear said, it's actually extreme clipping. The visualization window shows red and then sound dissapears. On some lower LFO amount values sound even comes back and then loses again and so on. I can lower the amount of LFO to get sound completely back instead of reopening the project.

softrabbit commented 10 years ago

The bug seems pretty old, 0.4.15 behaves the same. Off the top of my head, could this be filter calculations going crazy when the cutoff frequency goes too high? The RC filters seem pretty immune, but others break around 8000 Hz with full envelope and LFO modulation of cutoff.

diizy commented 10 years ago

On 03/18/2014 07:08 PM, Raine M. Ekman wrote:

The bug seems pretty old, 0.4.15 behaves the same. Off the top of my head, could this be filter calculations going crazy when the cutoff frequency goes too high? The RC filters seem pretty immune, but others break around 8000 Hz with full envelope and LFO modulation of cutoff.

—

I can't reproduce this bug. Could this be dependent on the audio backend?

softrabbit commented 10 years ago

Dependent on sample rate more likely. Clamping the frequency to m_sampleRate/2 in basic_filters.h seems to fix the problem, but it might change the sound in some cases.

diizy commented 10 years ago

On 03/18/2014 08:38 PM, Raine M. Ekman wrote:

Dependent on sample rate more likely. Clamping the frequency to m_sampleRate/2 in basic_filters.h seems to fix the problem, but it might change the sound in some cases.

—

Eh, let's clamp it anyway. Most people don't compose music for dogs ;)