Diffuser Chorus

[Tim Kleinert]

Here's a diffusor chorus FX circuit based on quad cascaded allpass delays modulated by quadrature sine LFOs (again offsetted 45 degrees between L and R channel for a nice wide stereo effect). Goes from subtle chorus to super-lush almost reverb-like tails. (Which shouldn't be a surprise to those familiar with Schroeders reverb concepts.)

Controls:
DELAY Size: Absolute delay size of the allpass delays, from 0 to 200ms.
DIFFUSION Amount
MODULATION Amount: How much of the selected delay size is then modulated by the LFOs, from no to (almost) peak-to-peak modulation.
MODULATION Rate

The parameters interact with each other. Eg., a modulation rate setting which works well at a small delay size will have to be lowered when you raise the size. There was no DSP left for ganging these to each other (which isn't that trivial if you want to do it correctly), so use your ears. High diffusion amount settings can also cause clipping because of the fierce internal feedback of the diffusor circuits. I tweaked it to be as safe as possible. But if you still get clipping, lower your input signal a tad.

Fits on a DSP.

And a short demo showing how the simplest possible single-osc sawtooth pad is transformed into a almost symphonic experience.



cheers,
t

EDIT: Note: DO NOT change the module arrangement of the allpass delay cascade. The module calculation sequence is crucial here. The diffusion feedback loops won't work properly otherwise and will cause ugly clipping. Took me a minute to figure out how to trick the patch compiler into doing what I want...

---

DiffuserChorusTK.pch2
Description:	Chorus effect with diffusion based on quad cascaded allpass delays. From subtle chorusing to super-lush almost reverb-like FX.	Download
Filename:	DiffuserChorusTK.pch2
Filesize:	4.33 KB
Downloaded:	5064 Time(s)

---

---

DiffusionChorusTK_demo.mp3
Description:	Short demo of the diffusion chorus effect, transforming the simplest possible one-osc sawtooth pad into an almost symphonic experience	Download
Filename:	DiffusionChorusTK_demo.mp3
Filesize:	3.81 MB
Downloaded:	1659 Time(s)

---

[ian-s]

magnificent!

[Lfohead]

Thanks for sharing, keep the great work up mate

[Kaan]

Wow! Amazing... Thanks for sharing!

[Check Mate]

Thanks for that awesome patch Tim!

Cheers

[dorremifasol]

Wow, another gem. Thank you Tim! I'm trying it on the demo right now. Have you noticed that lowering the diffusion amount introduces a considerable delay?

[Tim Kleinert]

[iPassenger]

Feckin nice demo Tim!!

Not looked the patch yet but the demo mp3 had my jaw hit he floor!

Sick... reminds me of my eventide.. but cleaner sounding.. very very nice.

[dorremifasol]

[iPassenger]

TIM,

Any chance u could give us some idea as to how this all works and what these all pass delays are doing. I've tried making all pass filters and delays before to smear signals but not got that far in the main.

[Tim Kleinert]

As mentioned, the L and R channels both contain 4 allpass delay stages arranged in series.
The usual implementation of an allpass delay is a delay with feedforward and feedback, the feedforward being tapped after the summing of feedback with incoming signal. The problem with this kind arrangement is headroom. If you apply a continuous signal to it, together with high amount of feedback, clipping is imminent.
Therefore I use a variation of this design, which also injects the input signal, scaled by feedback amount and inverted, into the feedback loop after the delay stage, before recursion. The result is that the more the feedback is raised (=dispersion) the less input signal reaches the feedback loop (as it is progressively cancelled out by the loopback injected negative signal) which thus prevents clipping. At the same time, this configuration will therefore pass the original signal through increasingly as feedback is raised (and the feedback loop shuts itself off progressively). (This is also the reason why the module placement should not be changed, because otherwise the cancelling doesn't work anymore due to one sample of latency.)

The problem with allpass delays is that typical ugly metallic ringing. In a standard Schoeder reverb type arrangement, one tries to alleviate this by cascading multiple allpass stages and making their delay times mutually prime. Another way to prevent metallic ringing is modulation. Although this is frowned on by reverb purists trying to model a real acoustic space, personally I love the spacy artificial sound of modulated reverbs, especially on synths. Modulated diffusor delays are in fact the building blocks of these kinds of reverb algorithms.

In this patch, the 8 diffusor delays all share the same delay size, but are modulated by one single sine LFO, multiplexed into 8 outputs of 0, 45, 90, 135, 180, 225, 270, and 315 degrees (and smoothed to audiorate via integration). 0, 90, 180, 270 outputs modulate the L diffusors, and 45, 135, 225, 315 going to R, to sufficiently decorrelate the stereo image.

That's all there is to it. Hope this helps.

cheers,
t

[iPassenger]

That's great thanks.

So as long as I contruct the all pass delay the way you have with the modules laid out in that order it will operate correctly?

[Tim Kleinert]

Here's an isolated single diffusor stage, to show you what's happening. (I used a single 100ms delay though.)

In my patch, there's four of them each for L and R channels, in series and modulated as described above. That's basically all that's going on.

---

diffuserdelay.pch2
Description:		Download
Filename:	diffuserdelay.pch2
Filesize:	1.15 KB
Downloaded:	4159 Time(s)

---

[iPassenger]

Nice one, thanks Tim.

[Tim Kleinert]

Oh and I forgot to mention: As you see, the delay module is arkwardly placed above and outside the feedback loop. Don't change this! Otherwise the negative injection thing mentioned above will not work properly anymore. (Took me some time to figure that one out... )

[iPassenger]

Ahh this is great, thanks for this Tim... must be closely related to the eventide's method of verb production.

---

Dif Verb.pch2
Description:		Download
Filename:	Dif Verb.pch2
Filesize:	4.32 KB
Downloaded:	4241 Time(s)

---

[Tim Kleinert]

[iPassenger]

[Tim Kleinert]

[jamos]

Diffuser chorus? What a boring name. You should call this the Hallelujah chorus!

[Tim Kleinert]