How to explain an analog state variable filter ..

[Tony Hardie-Bick]

On 13/10/11 10:04, Joost Schuttelaar wrote:
> On Oct 13, 2011, at 9:57 , Marc Nostromo [M-.-n] wrote:
>
>> Very nice indeed.
>>
>> Pretty funny because I was just thinking about implementing a state
>> variable filter this morning. Truth is I'm looking for a generic/not too
>> expensive filter whose source code is easy to find (mostly for quick tests
>> and benchmarking). And that I wouldn't spend 4 days coding :)
>>
>> Would that be a good pick ? any other suggestion ?
>>
>> http://www.musicdsp.org/archive.php?classid=3#142
>
> I guess this one will sound like the VSTs of 1998 :)
>

I used Chamberlin's SV code when I was a newb, and ran it on a 56K DSP. It
sounded awesome on white and pink noise, but even with a non-linearity (on the
LP integrator loop, in the feedback going to its input, not affecting the output
directly: change "D2 = L" to "D2 = f(L)" IIRC - there you go Marc ;) ) it still
sounded digital on pitched waveforms. Not sure why this is - possibly cos I
didn't model the component noise.

Would like to do a good SV at some stage (ie work with a real circuit and get
them matched up). It sounded incredible for filtered noise (space ship rumble,
wind sounds etc.)

Oh yeah - I used two of them, cascaded (for 24dB/oct), with the first having
slightly lower resonance, and adjusted the resonance ratio by ear to give it
"bounce" just below self-oscillation of the second one. It was pretty killer,
but it didn't sound analogue, even with all the phase-locking weirdness and 
self-oscillation... ;)

Tony (HB)