Recreating Soundtoys Echoboy in Reaktor

srs

Has anyone tried to or know of any resources to help with creating an Echoboy style delay processor in Reaktor?

The specific character of Echoboy I'm interested in is the feedback algorithm that can produce some really interesting chaotic sounds that are more like an experimental granular engine.

Guessing the Tape-ish Delay in Core is a good start but there don't seem to be any variations on it in the user library...

colB

You need to be more specific. What do you mean by "feedback algorithm"? What is the feature of the feedback that you like that isn't in other delays?

I've made some delays, but it's hard to know if they are the sort of thing you mean, or completely different! There are so many different ways you can go with a delay design.

https://www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/7525/

https://www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/3603/

https://www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/6015/

Each of these has it's own unique personality, and you can get chaotic with any of them ;)

There are loads of other cool delays in the UL too!

Rob Gee

There is this in the user library, apparently it’s a cross between Echoboy and Primal Tap.

https://www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/9375/

srs

Thanks both for these recommendations. Lots to experiment with and study here.

colB - to be more specific, it's the way the feedback algorithm in Echoboy produces sharp chaotic bursts of the original audio when pushed to its max setting. It stops being a conventional delay and becomes more like a granular stutter sound. Some presets which explore this are the Richard Devine ones from Echoboy v4 (the 'Ae' ones in particular).

I think a part of this is due to the dc offset produced in the feedback resulting in scattered bits of silence which transition to sharp audible bursts of delayed 'grains' of the input audio.

I haven't found any modules that seem capable of this kind of behaviour or if Reaktor permits these weird dc offset artefacts in feedback chains? They might be ironed out somewhere in the underlying code for safety purposes..

colB

Seems like very odd behaviour for a delay.

Maybe some some quirk of the Echoboy implementation, and they liked the sound, so decided it was a feature rather than a bug?

uksnowy

https://community.native-instruments.com/discussion/comment/11488#Comment_11488

out of interest. What did you use to capture those graphs... ?

srs

Yeah I've got a thread going with Soundtoys about it as this bug/feature was removed in the last v4 versions and the v5 updates that followed.

I was hoping that Reaktor Core had some potential to produce interesting feedback delays that behave like this after experimenting with the Tape-ish Delay macro but no luck yet...

srs

https://community.native-instruments.com/discussion/comment/11563#Comment_11563

Adobe Audition. It's the Spectral Frequency Display

colB

https://community.native-instruments.com/discussion/comment/11577#Comment_11577

experimenting with the Tape-ish Delay macro

The tape-ish delay macro is more ish than tape.

There are two ways of changing the delay time on a delay. You can change the read position in relation to the write positin within the recording media, or you can change the update speed of the recording media.

BBD delays change the update speed via the clock used to drive the BBD chip.

Most digital delays change the read position - all the Reaktor factory delays do this, including the Tape-ish delay

Tape delays can work in both ways, some have a speed knob to change the transport speed of the tape. Others (e.g. echoplex) allow you to physically move the playback head - or let you switch between different play heads.

The issue here though is that tape is a mechanical device, and there are small changes in playback speed due to friction etc. that cause a chorusing or thickening effect. The only way to accurately simulate this is to use a system where you can change the update speed - modulating the read position only gets you half way there.

Echoboy claims to do BBD and Tape style delay, so I would imagine (maybe wrong here?) that it correctly simulates changes in the update speed of the whole system. You can't do that with tape-ish delay.

This is pretty much the reason for the delays I uploaded to the UL - trying to find a way to get closer to this characteristic of BBD and Tape delays that differentiates them from most digital delays.

All that said, it really doesn't get you a whole lot closer to simulating bugs in Echoboy :) For that you would need a whole lot more deatiled analysis, and maybe some inside information as to what causes the glitches. Then you might get closer.

Although with this kind of problem, being somewhat non-linear, your probably going to need very precise and detailed information to siulate it. Other option is to just go for something which a somewhat similar character. So, in great detail, what it it about the character of the effect that is important to you (Other than 'Richard Devine')?

srs

The specific character of the effect is made by the chaotic dc offsets that the feedback produces and the sharp transitions to and from silence.

The modulation of the read/write speed is likely what introduces the chaotic behaviour so it seems the key challenge is getting the feedback to overload the tape buffer to create clipping and dc offset artefacts.

Time to get deeper into building basic delays in core and experiment with the feedback chain at as low a level as possible...

ehdyn

Do you have an audio example you can point to.. Reaktor excels at this sort of thing.

srs

Here are some samples that demonstrate what I'm referring to. These were all made in that older version of Echoboy v4 (prior to the dc offset bug fix) and the source material was just generic whoosh/hit sounds like those produced by Tonsturm Whoosh.

https://www.dropbox.com/sh/6x7lga28vxey61h/AABSWngM0a2hFHMW0D81RosYa?dl=0

colB

So can you not just use some click generator for a similar suond, maybe with some filtering. The 'geiger' primary module can generate some clicks like that.

Is there a particular reason why this must be generated by a delay bug?

Note also, that generating large DC offsets and sharp discontinuities is not particularly good for equipment, speakers headphones etc. Should be OK as long as there is always AC coupling somewhere in the chain, but still... best avoided.

colB

Here's a hacked together glitch delay using the basic library delay, so up to 1 second of delay at 44100Hz.

I hacked in a DC offset that grows when feedback is at a very high setting. fiddling with feedback setting and delay time will cause glitches more or less, lots of hard clipping.

Unpredictable, so watch your output levels :)

glitch delay.zip

srs

Awesome. This is definitely doing something that is exactly like the behaviour I'm searching for so there is a lot to study in the glitch delay you made.

I hear you on the click generator approach but with feedback delays like this it's possible get those fragments of the original audio bursting through. In terms of the danger to equipment, for me this is more about using the results as source material in a spectral sound design chain. This stuff does weird things when put through FFT math as the clicks have maximum energy across all frequencies.

Thanks for all your help!

colB

No worries. I found that it can settle to different states, so if it settles to just a wall of clipped distorted noise, then you turn down the feedback, then abruptly throw it to max, it can go into clicky mode, or somewhere inbetween, or back to the wall of noise. All good fun anyway :)