This became one of my favorite techniques after I saw Jakub’s video over 1 year ago. Jakub does not use this term, but I call it recursive filtering. I’ve used the technique in many patches.
Yep - recursive filtering is a prime driving force behind my development of the Recurse module.
Yes indeed - F2 is a great choice for this technique, as it hones in on 1 or a small cluster of harmonic partials using very little CV, yet also manages to add some character.
I’ve been meaning to publish a patch that compares the suitability of various filters for this technique. I’ve had various forms for over 1 year! It deserves a demo video, but could never get around to creating one. I will try to post the patch without a video today so people can try it out.
Once my plugin is finally released to the library, I will publish another version of the patch that uses my plugin.
Yep - My Venom plugin has a pair of Recurse modules, one mono, and one stereo, that is designed to do this very thing. Venom Modules is in beta now, and I hope to release to the library soon (good luck defining what that word means).
Well, sure, a lot of filters using 4 way SIMD. As you can see the Surge, VCV, and F2 are all pretty efficient when used polyphonically. Now if you want to see one that is an absolute pig, my Stairway is awful! (mind you liquid filter is similar to Stairway).
Stairway is super slow because a) it’s really old. b) it doesn’t use SIMD. c) it’s 4X oversampled d) it’s a ladder type filter, but e) it has a selectable non-linearity in between each stage. It also has way too many controls on it, but they don’t add much CPU. It’s all in the manual, but it was supposed to be a mix of the matrix-12 filter, minimoog filter, the Rossum eu filter, etc… It didn’t end up sounding much like any of those but it’s pretty cool as its own thing.
If you are super interested… hey, wait, don’t go away! … the manual is here
update: I went and re-read that manual. It’s actually interesting (to me) and cites a lot of other filters that did similar things. And, since I wrote most of it, it’s still has some typos.
Oh, and another old article about using the performance meters is here
btw, this is a perfect example of what I tell ppl all the time - the range of CPU uses by a module can vary a huge amount from one to another. rather than worry about getting some 2X out of your setup, you can often get 10X just by picking a different module.