New filter/effect concept ?... Looking for technical opinions for constructing test experiments.

horsetrainer

I've been studying virtual/analog synth design with a current focus on identifying and categorizing the known scope of parameter modulation techniques relative to the particular sound characteristics each generates.

I'm finding it interesting that "apparently" much of the sound-result of modulation types, can be understood (viewed) as methodologies for selective introduction of pitch-shift elements into a primary signal.

For example... Each side of a pulse wave can be viewed as it's own waveform. The sound effect of modulating pulse width occurs because one side of the pulse wave is pitch shifted relative to the other side. A single PW-modulated square wave oscillator can be viewed as one generator creating two waveforms. One waveform of stable pitch, the other of a varying pitch.

....

Now I'm investigating filters, and the complex math is where I'm getting stuck.

I have a basic understanding of theoretical (perfect) filter function.

But what I'm wondering is if imperfect Analog circuit designs, might intentionally or unintentionally be capable of introducing imperfect harmonic frequencies together with perfect harmonics?

Now I'm wondering if "selective filter harmonic detuning" is already an existing effect... And if not how I might be able to experiment with it ?

To reiterate.. Filters are designed to attenuate and/or amplify specific harmonics in a primary signal. The result is the filter behavior we are all familiar with.

What I'm saying... Is to modify that traditional filter functionality, by adding the ability to selectively vary the pitch of single harmonics, so they are no longer true harmonics of the primary signal.

I'm guessing this would most likely have to be done digitally.

I'm not saying shift all harmonics out of tune. Just select harmonics.

Most importantly of all. This "selective filter harmonic detuning" must have the ability to apply modulation to selected harmonics to cycle then in and out of "a state of being a perfect harmonic".

rs2000

It's certainly worth to learn what filters do in the time domain and in the frequency domain. It's also worth a lot to understand what different filters, especially ideal vs imperfect analog filters do.
Shifting select harmonics is nothing a classic analog filter can do, that would be an artificial digital design operating in the frequency domain (Signal > FFT > Harmonics tweaking > iFFT > Processed signal).

What differentiates the various analog filters is not only the circuit design but also the components used. Each introduces its own nonlinearities that cause harmonics not existent in the original signal and the art of combining different electronic components into a unique filter circuit is not purely something you can do on paper, it's often a lot of trial and error until you have something that actually sounds really good.

Check out @Krupa's tube circuit experiments, they're a great example of how to make use of this fascinating area of physics and electronics.

horsetrainer

@rs2000 said:
It's certainly worth to learn what filters do in the time domain and in the frequency domain. It's also worth a lot to understand what different filters, especially ideal vs imperfect analog filters do.
Shifting select harmonics is nothing a classic analog filter can do, that would be an artificial digital design operating in the frequency domain (Signal > FFT > Harmonics tweaking > iFFT > Processed signal).

What differentiates the various analog filters is not only the circuit design but also the components used. Each introduces its own nonlinearities that cause harmonics not existent in the original signal and the art of combining different electronic components into a unique filter circuit is not purely something you can do on paper, it's often a lot of trial and error until you have something that actually sounds really good.

Check out @Krupa's tube circuit experiments, they're a great example of how to make use of this fascinating area of physics and electronics.

Thanks.

I'm doing experiments with a free desktop wavetable synth called Vital at the moment. Just to explore the harmonic detuning aspect of the concept. Vital has a wavetable creation tool where you can adjust harmonics one at a time.

I set up two oscillators in Vital. One with a saw wave.. The other I pull up just a couple of harmonics (like organ drawbars). Then I sightly detune the harmonic OSC against the saw wave OSC. The result is a detune effect that emphasizes the presence of the detuned harmonics.

Finding a way to selectively detune a filter response could be a challenge using regular apps..

But I like the concept of a filter that can apply a sort of frequency targeted detuned unison resonance effect as a part of the filter response to the input signal (or by ext. modulation).

I'm not sure. But I don't think analog circuits can do this because they use feedback to excite resonant stages. Shifting the frequencies of the analog circuit seems like removing the essential frequencies it needs to produce resonance.

I think a digital version would be cool. Because it (hypothetically) could be programmed to provide filter response in various "unnatural" ways.

NeonSilicon

A digital filter isn't going to be able to do what you've described either. They are basically stuck with the same limitations as an analog filter for this sort of application.

Depending on how precise you want the pitch shifting to be, there are some things you can try.

One thing you could try to see how the effect works would be to do some modulated pitch shifting of individual elements from an additive synth.

For a generic effect on any given signal, if you wanted to pitch shift whole ranges of harmonics, then you could combine band-pass filters with vibrato sections to build up a set of pitch shifted regions that you could then recombine. This is going to get more complicated if you want the modulation to precisely move the pitch of the sections. But, you could could still do something that would amount to multi-band pitch shifting. Maybe split the signal up into regions and then apply a modulated autotune to individual regions.

If you want to do something like pull out the third harmonic of the input signal and then do a modulated pitch shift of that, I think you are going to have to go into the frequency domain and do the manipulations there. It's going to get complicated too.

If you are synthesizing the signal yourself, then there's more you could try. You could do something close by modulating the signals in FM in certain ways to drive the inharmonic elements.

I think the place I'd start would be with the additive synthesis approach. That way, you could easily start with a base set of partials that give you the waveform you want and then modulate the pitch of a particular set of partials and see what it sounds like.

Samu

I got a bit 'Virsyn Vibes' from the idea for some strange reason...
Varying the overtones is in practice adjusting the level of the individual partials in an additive synth.

'Prototyping' the idea in say for example Drambo could be worth a shot...
(Ie. create a bunch of sine oscillators and use a a pair of graphical modulators to control the amplitude and frequency of each oscillator and mix them together).

horsetrainer

@NeonSilicon said:
A digital filter isn't going to be able to do what you've described either. They are basically stuck with the same limitations as an analog filter for this sort of application.

Depending on how precise you want the pitch shifting to be, there are some things you can try.

One thing you could try to see how the effect works would be to do some modulated pitch shifting of individual elements from an additive synth.

For a generic effect on any given signal, if you wanted to pitch shift whole ranges of harmonics, then you could combine band-pass filters with vibrato sections to build up a set of pitch shifted regions that you could then recombine. This is going to get more complicated if you want the modulation to precisely move the pitch of the sections. But, you could could still do something that would amount to multi-band pitch shifting. Maybe split the signal up into regions and then apply a modulated autotune to individual regions.

If you want to do something like pull out the third harmonic of the input signal and then do a modulated pitch shift of that, I think you are going to have to go into the frequency domain and do the manipulations there. It's going to get complicated too.

If you are synthesizing the signal yourself, then there's more you could try. You could do something close by modulating the signals in FM in certain ways to drive the inharmonic elements.

I think the place I'd start would be with the additive synthesis approach. That way, you could easily start with a base set of partials that give you the waveform you want and then modulate the pitch of a particular set of partials and see what it sounds like.

Thanks for your assessment on how to investigate further.

I'm going to start with an additive approach and see what I can learn.

I like these kinds of experimental challenges. They often lead to unexpected discoveries.

horsetrainer

@Samu said:
I got a bit 'Virsyn Vibes' from the idea for some strange reason...
Varying the overtones is in practice adjusting the level of the individual partials in an additive synth.

'Prototyping' the idea in say for example Drambo could be worth a shot...
(Ie. create a bunch of sine oscillators and use a a pair of graphical modulators to control the amplitude and frequency of each oscillator and mix them together).

I'll take deeper look into the Virsyn apps. I do have AddStation already, I'll need to take some time to learn exactly what I can do with it.

I'm thinking of creating some experimental simple harmonic wavetables in Vital on desktop, then moving them into Drambo's wavetable Osc for further experiments.