Modal Resynth Feedback Problem

Hi! I took this EasyFFT resynthesis ensemble by Jan Brähler and replaced the sine bank with a modal bank to do a vocoder like thing

https://www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/6729/

But there is a problem
When the resynthesized sound is dynamic, if you set a long delay, it seems like it starts to feedback a lot until it reaches a full distortion crackling sound

Resynator Test.ens

Also it works only with L as input and R as sidechain, couldn't manage to make it stereo

Find more posts tagged with

Reaktor 6

Reaktor Building

Comments

ANDREW221231

this sort of makes sense if you think about it: the modal bank can work with dynamic live input but with a long delay time each slowly decaying resonation will stick around and eventually accumulate

kind of like if you set the feedback of a delay to 1 and played guitar into it, eventually it will build up into an atonal mess

Frasco

@ANDREW221231 Yes, but at the same time I notice how, increasing the smoothing parameter or freezing the resynthesis doesn't cause the distortion even with extreme decay times

That's why i was thinking of something related to the dynamics of the sound

Also made a EzFFT 2048 version and doesnt have this problem

Tr97

In your ensemble the input signal is used as carrier (R, exciter) and modulator (L, extracting amplitudes) - this is analogue to a vocoder where a signal more or less vocodes itself - this could cause the extreme impact of loud passages. You could replace the exciter by a single constant noise or an oscillator like done with vocoders to achieve different results.

A decay value of 80 is 10 seconds, maybe this is a bit long. The used iteration reduces higher values very quick - this results in an extreme long decay for the bass frequencies and almost none in the higher ones, this also could impact the sound.

Frasco

@Tr97 For solving this in Ableton I've routed both L channels of exciter and modulator as L R in one Reaktor instance

And did the same for the R channels

If you mean that Iteration, it's the damp added directly in the decay data

Maybe you're right, the fast fluctuation of the bass frequencies could be the problem, maybe attenuating them might work? Idk

Quick links

Unanswered questions