[.uhm] Karplus-Strong in Hive

[fmr]

Might start the FM-Turorial today

Please post a link in the Tech Preview thread, or in this one

[justin3am]

Incredible stuff, Urs! I've been playing with .uhm scripts all day (nothing worth sharing).
The concept kind of blew my mind. At first I was like "well it sounds cool but Hive doesn't have patch cables so..." but I've had a lot of fun just tweaking the example scripts. I think the fact that I'm actually learning something in the process makes it extra fun/rewarding.

[ZeePok]

 
Great stuff, Urs! Thanks!

Since these [.uhm] threads are major Tutorial stuff, could you please make them sticky?

I will post a few more tutorials and then make a general Sticky with links to each.

Might start the FM-Turorial today

 
Thanks!

And thank you for doing them in text form. I found that I study much better from texts than video's

[hakey]

In the SimpleKS example can we replace the 6db lowpass with tuned resonant bandpass filters in parallel? If so, how?

[Urs]

In the SimpleKS example can we replace the 6db lowpass with tuned resonant bandpass filters in parallel? If so, how?

Yes, but it is not trivial.

Code: Select all

Info "Simple Karplus-Strong-like wavetable. Use Auto and adjust to taste."
NumFrames = 101
Seed = 1346
Wave end=0 "rand * 2 - 1"
Wave start=1 "0.96 * (0.95 * main_fi(frame-1, index) + 0.2 * bandpass( main_fi(frame-1, index), 0.0, 0.0) + 0.1 * bandpass( main_fi(frame-1, index), 0.2, 0.0) + 0.1 * bandpass( main_fi(frame-1, index), 0.5, 0.0))"
Normalize base=each

Unless you constantly feed noise, bandpass filters either die quickly or build up a takeover.

The problem is to set the overall feedback as well as the individual amounts for dry signal and added in bandpasses so that something useful comes out in the end. Setting resonance makes this nearly impossible.

As a workflow hint: Use Normalize base=all as a starting point, then balance things out (tedious!), then maybe go back to Normalize base=each, or keep all.

Note that I'm considering formant synthesis as next tutorial, if/once I find the time to. Instead of Karplus-Strong, I'd explore noise/exciter generation and bandpass filtering for those vowel-type wavetables.

[hakey]

Instead of Karplus-Strong, I'd explore noise/exciter generation and bandpass filtering

Yes, I was rather thinking along the lines of modal synthesis for simple mallets and the like.

But I guess that it's more involved than KS with bandpass filters.

[Urs]

Instead of Karplus-Strong, I'd explore noise/exciter generation and bandpass filtering

Yes, I was rather thinking along the lines of modal synthesis for simple mallets and the like.

But I guess that it's more involved than KS with bandpass filters.

Yeah, probably.

It's difficult to distinguish between modal synthesis and additive synthesis in respect to wavetables. By nature, wavetables can't represent dissonant frequency components. Modal synthesis with harmonic exciters is thus pretty identical to additive synthesis by result. But of course one can use noise and stuff with resonant bandpass filters instead.

I guess I'm going to go there shortly, but I also want to figure out a visual guide to how it all works.

[hakey]

Modal synthesis with harmonic exciters is thus pretty identical to additive synthesis by result.

Okay, I see what you're saying.

[fmr]

So, what about that FM Tutorial?

[Urs]

So, what about that FM Tutorial?

viewtopic.php?f=31&t=511639

The things people do are actually so advanced, I'm not really sure I'd be the better tutor for any future threads. I'll come up with something... just a bit busy now.

[surreal]

In the SimpleKS example can we replace the 6db lowpass with tuned resonant bandpass filters in parallel? If so, how?

Yes, but it is not trivial.

Code: Select all

Info "Simple Karplus-Strong-like wavetable. Use Auto and adjust to taste."
NumFrames = 101
Seed = 1346
Wave end=0 "rand * 2 - 1"
Wave start=1 "0.96 * (0.95 * main_fi(frame-1, index) + 0.2 * bandpass( main_fi(frame-1, index), 0.0, 0.0) + 0.1 * bandpass( main_fi(frame-1, index), 0.2, 0.0) + 0.1 * bandpass( main_fi(frame-1, index), 0.5, 0.0))"
Normalize base=each

Unless you constantly feed noise, bandpass filters either die quickly or build up a takeover.

The problem is to set the overall feedback as well as the individual amounts for dry signal and added in bandpasses so that something useful comes out in the end. Setting resonance makes this nearly impossible.

As a workflow hint: Use Normalize base=all as a starting point, then balance things out (tedious!), then maybe go back to Normalize base=each, or keep all.

Note that I'm considering formant synthesis as next tutorial, if/once I find the time to. Instead of Karplus-Strong, I'd explore noise/exciter generation and bandpass filtering for those vowel-type wavetables.

That would be cool!

[lengro]

EDIT: (deleted - found answer in "[.uhm] FM Synthesis in Hive" thread)

[surreal]

following up 2021