lmms
lmms copied to clipboard
LMMS
Dual filter highpass problem
I'm having problem with the dual filter effect, what should i do? When i activate an highpass on the master at 1hz at some point of the track with an automation, it does a strange thing, like volume turn down and the highpass is not at 1hz. On dual filter, the mix button is totally to the left (so the highpass filter), and i have automations to decide the frequency, too. My automations chain is 'highpass activated (start at 1hz)', 'frequency go up', 'highpass disactivated', 'turn frequency to 1hz' and so on, for multiple times. Like in the image below.
The first time the highpass is activated nothing happens, but from the second one happened the thing i said before. You can hear it here in a snippet on the track i'm working on at like 01sec: https://soundcloud.com/h4ckton3/a-1/s-ITbUR
It happen also on other parts of the track anytime the highpass is reactivated. I also post here an image of the waveform at that point if it could help (you can see it's weird).
What it could be? I guess maybe is the effect that answer too slow to the automation control (i hope you can get what i mean, i'm not english).
I really need to fix that, because my track is nearly finished and that's the only problem it has. Thanks
Installing RC7 didn't solve it. I can avoid it by using several highpass filters on the master, but it's only turning around it. If someone will manage to fix it, please tell me
Have you tried easing the automation back down instead of having it change suddenly? That might have something to do with it.
I don't reall get what you're saying (i'm not english), but as you can see in my automation chain, cutoff return to 1hz while the highpass is not activated, and only then i activate the highpass and cutoff turn gradually up,... Also this is how i wan the track to be to so i don't want cutoff to return down bef the drop.
P.S: Maybe i haven't understood well, if so, try explain it to me better, please
I really hope this will be solved, because i've just remastered the track i had this problem on, and after making all sounds good, even if i use 5 different instances of dual filter on the master, i can't totally solve the problem, and so i can't publish it. Also mantaining highpass filters always activated even with cutoff at 1 will make the track losing some bass frequencies.
@RoxasKH Can you submit a minimal project that shows this bug/behavior? (it need to be in .zip format to be dropped in a github comment)
The last two automations tracks are the Highpass Filter ones, and they are the ones that cause the problem
What I experience is that when a filter rapidly has it's cutoff dropped to near-zero, it resonates like that for a few samples.
@RoxasKH I couldn't work with that large project and didn't see what you meant with it. I would appreciate a minimal version of that one to demonstrate just the bug and nothing else.
I tried and create a minimal project with just the described components and this is what I came up with:dualfilter.mmp.zip It's a super beat with a high pass filter under automation and which has it's bypass controlled by an lfo. You just hit play and leave it looping away and there will be glitches in there eventually. It may be the same thing you're describing, in which case it's not a floating point error as it passed just fine under a debugger.
@zonkmachine I massively reduced the project, tell if it's still too much and i'm gonna recreate it even littler. It wasn't so noticeable in your project so i'm not sure if it's the same thing.
This is a bit like plugging and unplugging wires, which can give you awful pops and clicks on the output if your gear is on.
I believe the problem is that when the filter is bypassed, it won't be fed any audio. That might leave it in an less ideal state as the output depends on what has come before, and the stuff before is whatever there was when the filter was disabled. Am I making any sense here?
I'd suggest automating the wet/dry knob instead as a workaround.
The problem is fixed by pull request #7237 which extends the Effect base class with a new virtual method that can be overridden to react to state changes with regards to being enabled/bypassed.
Currently only DualFilterEffect overrides this method by resetting it's two filters to get rid of the pops described in this issue.
While testing this I got one segmentation fault in Effect::isRunning but i was not able to reproduce it. So I would appreciate if this could be thoroughly tested.
