The per voice modulation nonlinear curve problem

[Faiky]

Hi there,

I've been asking myself this for quite some time, but why are the voice modulations not linear?

Per-Voice-Mod-1.jpg

I have to compensate for them all the time, but I can't get it exact.
When the knob is at
50%, the value is 25%
25% is 6,3%
75% is 56,3%.
The curve to compensate looks like this:

Per-Voice-Mod-2.jpg

I wonder if there is a mathematical background to this, but there is no predetermined curve to compensate for this exactly.
But it would be nice if you could switch it to linear, so that 50% is 50%, 25% 25% and so on.
I think this would also save unnecessary calculation for the cpu.

If it's a small thing, I would be happy if you can switch to linear in the future. Thank you and I'm very happy with MeldaProduction. Stay healthy!

[Faiky]

Oh, and could you very please add a "constant per note" modulator? So that when you modulate it, the value is always changed for the note only, and not changing in real time. Like if you modulate "Note Panorama" in the global tab.

If it isn't too much trouble, I can use it really well and it's always one thing I miss every time in instrument building.
Thank you thank you thanks in advance .

[MeldaProduction]

I don't follow - modulations ARE linear, but some parameters are NOT! Imagine frequencies - 20Hz to 20kHz, if it was linear, then in the middle there would be 10kHz , unusable. This one is logarithmic. Volumes are usually squared. Etc. Most stuff is linear though.

The modulator you are asking for is called Note random I think

[Faiky]

I mean what's in the first picture. A1 here. This value down there (22.1 on the picture) can be modulated. But at 50% A1 is on 25%.
You can also see that the blue indicator is almost halfway, but only shows the 21% value.

[] Peter:H []

Oh, and could you very please add a "constant per note" modulator? So that when you modulate it, the value is always changed for the note only, and not changing in real time. Like if you modulate "Note Panorama" in the global tab.

If it isn't too much trouble, I can use it really well and it's always one thing I miss every time in instrument building.
Thank you thank you thanks in advance .

I have asked for almost something to the same direction. In my case I 've asked for a counter that progress es with incoming "gate signals": when a note coming in, the counter would add +1 until a max value of x - configurable as well - is reached and the counter wraps around and starts over at 0 again. This would need true gate signals which is currently not present in MSF.
What was added with 14.05c beta was "Counter 4|8|16" per voice modulations, which count but not with a configurable max/wrap around value. With it I was able to sort of "simulate" per-note modulation with a trick - a intermediary LFO with almost zero frequency which is phase-modulated by the counter, see here: viewtopic.php?f=138&t=542732
Link contains a example patch. Note: Only works in 14.05c beta.

In my eyes MSF is kind of a hybride but not a true modular. Compared to Bitwigs "The Grid" the whole area of Gate Signal Processing is not comparable to true modular stuff. For instance I haven't found a delay for a LFO, i.e. a delay is kicking in only after some msec after the note gate... in Bitwig i would simply use a standard delay module and put it right before the gate-in of the LFO ...
The gate thing would also help the sample and hold module ... anyway we get pretty much insane other stuff in MSF and this complaining of mine is on a very high/nerdy level.

Stay healthy!

BR
] Peter:H [

[Faiky]

I don't follow - modulations ARE linear, but some parameters are NOT! Imagine frequencies - 20Hz to 20kHz, if it was linear, then in the middle there would be 10kHz , unusable. This one is logarithmic. Volumes are usually squared. Etc. Most stuff is linear though.

The modulator you are asking for is called Note random I think

I mean what's in the first picture. A1 here. This value down there (22.1 on the picture) can be modulated. But at 50% A1 is on 25%.
You can also see that the blue indicator is almost halfway, but only shows the 21% value.

I need to explain the problem more clearly. This parameter with 21.1 %, i.e. the modulation depth of e.g. Mod A Slot 1. This is the parameter I want to assign to an MP., which is no problem. But exactly this parameter always behaves recursively. No matter which modulation source or modulation target.
It's with each per voice modulation source.
Correcting with the transformation curve worsens the performance and is also not exact.

[DarkStar]

Sorry, I do not understand either Can you attach a Clipboard copy (using the Copy button in the Toolbar) of an example?

"When the knob is at
50%, the value is 25%
25% is 6,3%
75% is 56,3%."
To which knob and value are you referring?

I have just mapped an MP to a True Random Depth PV modulation and all seems OK. But i may have misunderstood.

[Faiky]

Same thing there too. I made a GIF:

[DarkStar]

Ah ha - I can now get similar results:

>>> https://i.imgur.com/uEWzZbz.png

Edit: Oops I got my columns and rows mixed up in the notes in the screenshots. For columns A, B, C and D, please read columns 1, 2, 3 and 4.

Here I used a MP (named "Master" and assigned another MP and 5 Per-voice modulator Depths to it. The ranges of the target parameters were -100% to 100% or 0% to 100% as appropriate. in many cases the figures displayed in the PV modulators Depth sliders seem incorrect (the length of the bars looks OK, I think). Either there is a display bug there or I have missed something obvious

To try it out, paste this code into a spare A...H slot in MSF

Code: Select all

$eNq9mF9z2jgQwN-5FBrltSm2bAOZsdxpKOQyDQmHadp7VPAGPBESI8m55D79jaQYCCGY3szlbbX89o9Wa3tF+uVpydEjKF1KQXH4OcAIxEwWpZhTXJn70x7+krXSUS4rUQzZzEj1rMGYUsw1OqE4xOjkgss7xvWU3bn1oCiNVHn5D1AcRt3ep07Y3dYOmH6mOAw6vU-dKMbojz9-rDRbrriLGWB0NclXvDQGHB5a1nKv1cTFGrGnq+HNOZezBx8wIhiN+tam44SAYhI6KaSYJE4iFJMzJ0UUx10nxRQnnksoTjzXoThxnDUIrBBRTGIrxBSTnhUSimMXycKO6VKcOKZHceKYM4ptBUYVN+WYKbYEA0r3JedspaFwe76tj+AsPCNnHZyleXU35tW8FC9Vz1K-zF9Xf38FZFFxprZChBjlwGFmoPBuKA4+Rbh2qmsBnTPx0JeVMM5mWHF+zZZA8Y2elZwzIxVyUfXDpXCZ7-6M0S+n-4ti69-6C9rvJH-yQ5RaioFgdxxeZ3tyK3llPQef3bHt2WaWjm6+neSwLI0UoNFL2gSjb7AyC+smSyfBll6rWUDxVFWAFBOFXDpVuKN6Cig+9WGfQptA1E1wO0snYe0qwkiuQorz6q52OmKlsJXx-tzKByQvq+gdx+iJbMVov9pSluYLtgKUTwfj8WSQD6autI+MV+7hchavK7tWrAW9j3nTXa2thvv1znHtO4Q9bRVs2mpvftuBWulXY9jsIczSgXgELleABte39WaJ22GN1DBphskajprhaA3HzXC8hpNmOFnDnWa4s4a7zXB3Dfea4V7WskyYpVMFS8nlvrayuIMcS45hiWejY9jIs-ExbOzZ5Bg28WznGLbj2e4xbNezvWNYW1--FRyIx0aDDdlKr4Y3IboCMbfvK-s07dCOII1E1EjEjUTSSHQaiW4j0TtMTNw72DUqE-peqiUzpRTbjd1zRazB2oQca0LWJtGxJtHaJD7WxHbwd3jWM8bhiN1s0I0ZOd7M7ukWuJyV5uiQO-yOA-KbDrYz2EVbG1djBVpXCg4hpZktzkEUvw5AI1n8XADwA8i0XN4p6MYHkHMFzCwCcgAZPK1syqUUYXgAyytt7Ef+feJaGri5v39LXPCyAP+qePNbXwqjJOegNMpLMefQXzAhgNePy85Mme0oAnTx47IvuVQUD4dnnSAYDrHV+Wl8JAs7R1-bM+ZuaoTNys91I6YNKIzyFdSTqh8+-BjjFt5uDGoGwmj78a-ju-lxwsQcKI4xWuu97wsQoNxEeYq2x0t0ilwToFO0Ge6WskDoFH0NNxnYLvw-Q5H-FMot7bT3QWmef1xFzsnHhdopYHunsXc7nWx1uu3yTvy7nd7nJQjz4Z3urzkHN0veaLQd-mamfAT7QgWD6rvTVIFz4q7b-gLUZwbmUj1fGlheinuJxkzrv6UqKLax+pVSIPZ48JlvOcgNM5ClQ2CmUqBdCHtvs2Fs0jtc+8Vzln4d+yzdPLqVtd2EWqEJM+DH1VY6crsjWfqzLMxilKW5rNQM0FjBfckNqJf-CKyrGmnXRq20ve9-Cp21-gWzJkSL

[Faiky]

It gets more wired.

[Faiky]

Okay, so the other curves (those from 0-100%, not those from -100 to +100%) seem to have a different bending. I tried it to compensate by a tranformation curve and it looks like this:

Per-Voice-Mod-3.jpg

It's kind of crazy and unexplainable to me why the bending point is at 50%. So 50% is 50%, but everything else is bent.

Thanks for the analysis, DarkStar, i hadn't noticed that before.

I've illustrated the whole thing using a triangle-shaped LFO:

modulation-curve.jpg

The first curve is from the first PV parameter, i.e. -100 to +100%, the second the other three, i.e. 0 to 100%.

[Tagirijus]

Never really noticed this, interesting. Maybe it has something to do with the mod going from -100% to +100% or so?

Here I tried to draw some line in the middle of the right sides of the values and 25% is in the middle, while 50% is at 3/4:

2020-07-31_11-31-06.png

Exciting.

[MeldaProduction]

Hehe so, the reason is that in most cases you need just a small amount of modulation. Full is really really a lot. So this "squared" curve focuses on low values and makes the workflow easier. Of course if you have OCD and need linear curve otherwise you get crazy, you are out of luck

Btw. these transformations are all around the plugins, simply because some parameters need it.

[Tagirijus]

Fair enough, hehe. Tbh, it doesn't bother me, luckily. But yeah: bummer for "OCD" people.

Thanks for explaining, Vojtech!

[DarkStar]

Understood. It would be useful (and prevent further queries) if that could be included in the Info page(s).

However, please have another look at my screenshot. The coloured bars displayed for the Depths do not match the numeric values. One of them must be wrong! For example, the first one has the MP value at 25%. In all the PV Modulator Depths the bar is 1/4 of the slider width; in which case the numeric values should be -50% and 25%.

Or, the numeric values are correct, based on the transformation that Vojtech mentioned. if so the bar lengths should be 1/8 and 3/8 of the width.