Soft clipper

[Logado]

i need create soft clipper (distortion)
The curve and harmonics are very similar to free plugin Gclip.
I found a formula that gives the right harmonics

Code: Select all

sample / pow(1 + pow(fabs(sample), alpha), 1.0f / alpha)

where alpha 2.5
but 2 power and 2 divisions per sample
Can there is another way to get this curve?
i try in desmos
https://www.desmos.com/calculator/ujkvajglhj
I did not succeed, can anyone have more experience in mathematics?

[TheOm]

That's very close to the good old tanh(x). I'm not familiar with that plugin but I assume that's what it's using since that function is very popular.
Now tanh itself is also expensive to compute, but you can get away with very rough polynomial approximations.
You can find some on google.

[martinvicanek]

There are fast stream functions over at FS Guru which you could use. Alternatively, try

Code: Select all

y = x/sqrt(1 + x^2)

which is even closer to your formula than tanh(x).

Edit: Sorry if this information was cursory. Here is a schematic with optimised transfer curves.

[Logado]

thanks for tips, distortion is an interesting area, so much magic happens here

[BobF]

Hello Logado,

DO some searches on the forum, but just not for distortion, but clipper, wave shapers, fuzz, square wave (etc), rectifiers, and so on. You will find many examples of magic. Feed them with a sine wave and then out to a windowed FFT display and you can view the harmonics.

Have fun.
Cheers, BobF.....

[Logado]

Hello Logado,

DO some searches on the forum, but just not for distortion, but clipper, wave shapers, fuzz, square wave (etc), rectifiers, and so on. You will find many examples of magic. Feed them with a sine wave and then out to a windowed FFT display and you can view the harmonics.

Have fun.
Cheers, BobF.....

I think I've seen ~90% of osm files from the forum I learned dsp with the help of this forum (thank you). Now I'm using the juce. I want to learn how to emulate different devices. But the smallest difference in the curve no longer gives the harmonics that need. I'm thinking of experimenting with a wave shaper with a table without formulas. I wonder what will come out