State Variable Filter for Synthmaker

lalo · 30 Jan 2005

amazing stephan!!! many thanks! lalo

stefan · 2 Feb 2005

triple-sampled version, stable up to nyquist Tunable without any sin1() call:

polyin input; polyin cutoff; polyin resonance;

polyout lowpass; polyout highpass; polyout bandpass; polyout notch; polyout peak; float rez, f; float lp1, hp1, bp1; float multi, x, x2, x3, x4, x5, x6, x7; multi = 1.0 / 3.0;

rez = 1.0 - resonance; cutoff = max(cutoff, 20.0 / 44100); // tunable version //f = 2 * sin1((3.141592 * (cutoff / 36.0))); // taylor series version of 2 * sin(x) x = 3.141592 * (cutoff / 36.0) * 2 * 3.141592; x2 = x*x; x3 = x2*x; x4 = x3*x; x5 = x4*x; x6 = x5*x; x7 = x6*x; f = 2.0 * (x - (x3 / 6.0) + (x5 / 120.0) - ( x7 / 5040.0));

lp1 = lp1 + f * bp1; hp1 = input - lp1 - rez * bp1; bp1 = f * hp1 + bp1; lowpass = lp1 * multi; highpass = hp1 * multi; bandpass = bp1 * multi;

lp1 = lp1 + f * bp1; hp1 = input - lp1 - rez * bp1; bp1 = f * hp1 + bp1; lowpass = lowpass + lp1 * multi; highpass = highpass + hp1 * multi; bandpass = bandpass + bp1 * multi;

lp1 = lp1 + f * bp1; hp1 = input - lp1 - rez * bp1; bp1 = f * hp1 + bp1; lowpass = lowpass + lp1 * multi; highpass = highpass + hp1 * multi; bandpass = bandpass + bp1 * multi;

peak = input + bandpass; notch = lowpass + highpass;

Dunno if the compiler atm does stuff like translating x / 40 to x * (1/40) etc, but I doubt it. I'll probably nag at jorge and malc about it anyway 🙂

lalo · 2 Feb 2005

great stefan! sorry if i ask again...can you do a little tutorial on the basics things that are happening math-wise... for example ..how can you obtain the 3x oversampling? i can't really understand it from the code....i think i lack some fundamental dsp concept about filtering

thanx lalo

stefan · 2 Feb 2005

As you may notice, a part of the code is repeated three times. That is the oversampling part. Also I correct the cutoff frequency for a sampling frequency three times as high compared to the normal one.

Btw, change x = 3.141592 * (cutoff / 36.0) * 2 * 3.141592; into x = 3.141592 * (cutoff / 38.0) * 2 * 3.141592; for a better tuning and frequency response. I will do the correct maths later to see what actually is the exact formula.

lalo · 2 Feb 2005

..argh what a dummy am i....i see now...so the calculation is done 3 times per sample..thanx lalo

malc · 3 Feb 2005

Thanks Stefan, great to see you trying things out.

lalo · 10 Feb 2005

hi stefan! is the 0..1 range of cutoff linear to the 0...22050 range? i ask you cause i'd like to have it mapped as log scale (am i right?) in other words as our ear perceive frequency/pitch.

thanx lalo

stefan · 10 Feb 2005

lalo wrote: hi stefan! is the 0..1 range of cutoff linear to the 0...22050 range? i ask you cause i'd like to have it mapped as log scale (am i right?) in other words as our ear perceive frequency/pitch.

thanx lalo

yes it is, I'd recommend that you for instance multiply the cutoff with 144 and then use a pitch to frequency converter to make it "linear for the ear" 🙂 Also, you can then add midi poly pitch to the cutoff (after you've multiplied the cutoff but before the conversion to frequency). It is tunable after all. Still haven't checked if the maths for the tuning is correct though, but I think they'll work just fine if you use the /38.0 version (see my earlier posts in this thread).

lalo · 11 Feb 2005

many thanx stefan 🙂

bManic · 2 Mar 2005

Just wanted to say thanks to stefan for his continued input and code!

Cheers! bManic

stefan · 3 Mar 2005

Just give me some more days and I will have something real to show you guys 🙂 Thanks bManic 🙂

lalo · 3 Mar 2005

stefancrs wrote: Just give me some more days and I will have something real to show you guys 🙂 Thanks bManic 🙂

😀 😀 watch out guys 🙂 thank you very much stefan!!

lalo

Tom7777 · 18 Mar 2005

Thanks 🙂