Analog filters, that sound "physical"

[itoa]

Yes, but this question is less important than possibilities of modelling other nonlinearities that seem more "correct".

[whyterabbyt]

Yes, but this question is less important than possibilities of modelling other nonlinearities that seem more "correct".

Then surely you're best modelling the 'thing', not filters that, when 'corrected' to better fit an idea of what the 'thing' is, still just approximate the 'thing'....

[itoa]

You are right. Unless one loves synths that sound physical... hmm cs-80 , tb 303, Buchla ? This matter is out of discussion. I think it's good to treat these things as unity and not stick to terms. Filters have certain aspects, everything is a filter: the circuit in the synth on my table and the table . At least what we do is physical modelling of electronic circuits

[whyterabbyt]

You are right. Unless one loves synths that sound physical... hmm cs-80 , tb 303, Buchla ? This matter is out of discussion.

Except that these synth filters are not 'corrected' the way you're talking about, yes? And that would mean that your 'corrections' are not what are responsible for making them sound physical.

I think it's good to treat these things as unity and not stick to terms. Filters have certain aspects, everything is a filter: the circuit in the synth on my table and the table .

I also think its good not to confusingly conflate terms, though.

At least what we do is physical modelling of electronic circuits

But you're saying that those electronic circuits are designed with 'ignorance'. You're not wanting to physically model real circuits, you're wanting to model hypothetical circuits idealised towards a behaviour which you think is 'more correct' for a specific purpose which the real circuits were never intended for. But you're quoting real circuits without those specific adaptations as examples able to produce the results you want.
Which is confusing.
For example you say "Why such a filters with physical behaviour are good? Well.. get a saxophone and blow stronger"

So what happens in that case that the correct application of modulators to amplitude, filter frequencies and resonance cant recreate?

[itoa]

Except that these synth filters are not 'corrected' the way you're talking about, yes? And that would mean that your 'corrections' are not what are responsible for making them sound physical.

Buchla's specific LPF gate mimics physical behaviour very well, both down-chirp and amp drop.
The same with 303 diode ladder + hardwired envs - the same effect is achieved, however it may be result of filter control.

But you're saying that those electronic circuits are designed with 'ignorance'. You're not wanting to physically model real circuits, you're wanting to model hypothetical circuits idealised towards a behaviour which you think is 'more correct' for a specific purpose which the real circuits were never intended for. But you're quoting real circuits without those specific adaptations as examples able to produce the results you want.
Which is confusing.
For example you say "Why such a filters with physical behaviour are good? Well.. get a saxophone and blow stronger"

So what happens in that case that the correct application of modulators to amplitude, filter frequencies and resonance cant recreate?

I mean that the opposite behaviour is worth checking. Maybe we can make use of nonlinearity that is more "natural" and a trumpet patch will start mimicking overblowing. Sorry if my loud-thinking is too chaotic. This is somehow how I see this for now

I will get back with some results soon.

[antto]

here's the x0xb0x filter with feedback level going far above 17.0
https://app.box.com/s/yh1z0dmcow63dgmyhlnv
the cutoff frequency itself is lower when resonance is zero, it goes up as feedback goes to 17.0, but from there, it doesn't really go anywhere (it audiably sounds like it goes a little bit down, but it's insignificant IMO)

[itoa]

Thanks a lot! How did you make this recording with constant cutoff?

So.. we have 2 filters that behave this way: synthi and tb-303. Suspicious: cs-80 and Buchla. Do we like them?

[antto]

well, easy.. when the filter envelope decays down - the cutoff freq is flat
i recorded the audio from before the VCA

whether i like it or not:
while playing with naive DSP filter algos (ladders with unit delay in the feedback) trying to figure out different ways to apply saturation, at different places, and different shapes, i saw a range of different effects
i saw detuning of the cutoff frequency in some setups as the feedback is increased beyond self-oscillation - and i don't like that tbh

[itoa]

I do the same again and again , just for curiosity. afaik in transistor ladder filters, the saturation should be located inside stages, just after difference of stage input and integrator output + one more for feedback difference. I'm not sure about other topologies, have to spent some time on electronics.

You should test it against input signal level (drive) as well. In the aspect we talk about feedback remains constant (material property).

[itoa]

I made the model of a filter with inverse (physical) saturation:
(ladder, bilinear integrated, 2x oversampled)

- SQR oscillator with noisy comparator
- A bit of distortion on the output

http://www.mixcloud.com/giku/filter-test-2/

The physical aspect of its sound is mostly emphasised on lower freqs.

btw. does this sound familiar?