Filter gain modulation not working correctly

[David Healey]

@Lindon My thought too. I've tried builds all the way back to 2023 and the issue is still there, so maybe this is a bug that has always been and I just never noticed - although I thought I'd used filters like this in the past without issue...

I'll see about implementing a fix.

[Christoph Hart]

@David-Healey afaik it was always like this it used the gain mode to modulate between zero position and the target value

[David Healey]

@Christoph-Hart

The dB readout for the modulator shows the correct value (I've increased it to 18dB here to make it clear) but that value doesn't make it to the filter. Maybe that's why I thought it used to work because I was seeing the correct number.

[David Healey]

Turned out to be a nice simple one, Claude fixed it.

[Christoph Hart]

@David-Healey yeah but doesn‘t that just break existing projects?

[David Healey]

@Christoph-Hart Might do... or fixes them :D

Preprocessor definition? Or another idea?

[David Healey]

Another little weirdness I noticed with the filter. Once a voice ends the filter value jumps to its previous value, this causes artifacts (clicks/pops) with other voices that are still playing.

[David Healey]

So after more poking around this issue with the filter values for voices overwriting each other also seemed to be present in the harmonic filter. It's very easy to recreate I don't know why it hasn't come up before, I think not many people are using voice start mods with poly filters or something.

I've updated the PR with the additional fix from Claude. In the process I also got it to fix a smoothing issue and a display issue.

[Christoph Hart]

@David-Healey alright, I checked your pull request, but it still is not sitting right.

I think we need to completely change that to use the combined modulation mode with a zero position of 0.5 - that's the convention for all modulated parameters that have a default position in the middle.

Then you can set per modulator scale / unipolar / bipolar and everything should work correctly. It's a pain to check and AI won't help us there as it's too brittle to explain to the robot (yay), but once it's setup properly it should handle all cases.

The default would be the scale modulation mode which should behave exactly as the pre-pull request non-bipolar mode (with the exception that the filter display matches the audio rendering). If you then want a modulation to go from -6db to +6dB, you just leave the knob position at 0dB and add a bipolar modulator with +6dB intensity.

[David Healey]

@Christoph-Hart Sounds good to me