Rev 0.2: General sound / efficiency improvements. Drift LFO now uses a chaotic equation like the pedal. A non-destructive panning LFO using the same principle has been introduced. The bitcrush effect is now applied only to the mid portion of the signal for a less extreme effect (which also sounds nice with the pan LFO turned on). Some not very useful controls have been removed. Feel free to leave feedback on these changes, and anything else you’d like to see.

Not the Cosmos, but perhaps a Cosmist. An alternative take on a Cosmos style looper, building on @the_burning_trestle’s excellent take.

I had two main goals for this: (1) to allow for smooth changes of the delay times without glitches or the doppler effect, and (2) to allow for long-term use of the blur effect without the dreaded frequency convergence effect which eventually turns everything into howling feedback. With a generous helping of crossfaders both have been achieved. Along the way I added some other features, but you don’t have to interact with them – page 1 is still all you need to start spinning out in the cosmos. See comments for a demo exploring many of the features via guitar and field recordings.

ARCHITECTURE:

The Cosmist consists of two sets of four delay lines of varying lengths, panned hard to left and right; the low C and D keys on the Organelle set the overall length, and increasing the ‘Drift’ control increases the distance between the four lines in each set. Changing the length initiates a crossfade between sets, avoiding an audible pitch change. The ‘Blur’ control determines the amount of feedback from the opposing stereo side into each delay line, resulting in a gradual blurring reverb-like effect that will eventually erode the original sound. A Blur setting of 100% simply reverses the stereo relation, so peak blurriness will actually occur at a setting of 50%. The ‘Smear’ effect ever so slightly randomizes the delay times for each line, adding a slight pitch-shifting effect and offsetting the tendency for high Blur settings to drift into ear-piercing feedback. The side effect of this is a slight loss of gain, which can generally be offset by boosting the Feedback control above 100%. The Smear effect is subtle from 1 to around 5 percent, but can get weird above this.

Those are the basics; I will describe the rest in detail in the Controls section, but in brief there are gain controls for monitoring the dry signal, loop volume, a gated reverb, and mix controls for lo-fi tape fx / distortion / bitcrusher. There are simple lo and hi pass filters at the end of the chain, and a suppression/compression circuit within the feedback chain for each delay line. There is also a chaotic LFO for the Drift parameter.

CONTROLS:

Keys: Use the ‘black’ keys for page navigation. Aux key feeds audio into the delay lines and is latching; footswitch does the same but is setup for my non-latching switch.
Low C: selects ‘short’ delay times of around 2.5 seconds at 0% Drift.
Low D: selects ‘long’ delay times of around 9 seconds at 0% Drift.
Low E: Leaves just two delay lines audible.
Low F: Leaves all four lines on.
Low G: Toggles between stereo and mono (left in only) input to the delays.
High B: Clear all delay lines.

All knobs correspond to their respective screen line. There is a knob-catching function as usual, so just move things back and forth if nothing happens initially.

Page 1: Basic looper controls, as described above.

Page 2: Gain levels, Tape FX mix. The reverb is a little experimental – I wanted to avoid continually feeding the delay lines into it, since things get blurry enough with this patch, so there are gates monitoring the levels of each delay line and opening an envelope after a certain threshold is met. This also avoids sending harsh onsets to the reverb. The result can be a very different soundscape than the loops, so their volumes can be controlled independently. The Tape mix determines the dry/wet mix for @audivit’s excellent generation loss port. At 50% you can get an interesting flanging effect as the wow and flutter deviates from the pitch of the dry audio.

Page 3: Wow/Flutter/Noise level controls for tape effect.

Page 4: Filter/Drive/Noise type controls for tape effect. Note that changing the filter or drive type will lead to momentary glitches in the audio, so set these to taste before use.

Page 5: Sample rate reduction mix/ratio, Low and High pass filter cutoffs. These are separate from and placed after the tape effects in the signal chain.

Page 6: Where things get weird! The suppression / compression effect works by switching the sidechain input for compressors in the feedback path. When the knob is at noon nothing happens; anything to the left and the current input determines the degree of compression, anything to the right and the current loop does. This is honestly the most confusing part of the patch and something I might work on more in the future, but it is capable of cool results. Just experiment with it a bunch while using different settings for the compressors (on page 7) and uses will hopefully become apparent. The suppression function can be used in a straightforward way to ‘replace’ audio in the loops, with the suppression percentage determining the amount of ducking and the response time the speed; the compression can be used to just smush everything that is already in there, as the experts say. Drift LFO: allows the drift setting to be randomized. Pan LFO: randomize stereo balance (should be roughly centered on average still).

Page 7: Compressor controls. For suppression a low make up-gain around 2 dB is usually good. For compression, a higher make-up gain is probably required to maintain unity-gain for the feedback path; also can sound good if things get crunchy.

Page 10 (high b-flat): Presets.

Let me know what you think! This isn’t intended as an improvement on the_burning_trestle’s take so much as alternate version, safer (to an extent which might be boring), but capable of different, more stable outcomes. And before anyone asks, I do not intend to add a reverse function, unless perhaps someone can explain how I might do that using the current delay line system rather than arrays. It would be cool, but I have always ended up running into audio glitches trying to do similar things in the past. Also a quick note on why I left the FX out of the feedback path: they turn everything to mush so quickly that it is not worth the hassle, especially since the managing the gain of the feedback path is already complicated. However, if using mixer sends for the input (like me), there is nothing stopping anyone from turning the dry level all the way down and recording the organelle’s output back into itself, resulting in deteriorating loops.

Acknowledgments:
@the_burning_trestle’s Cosmos adaptation: https://patchstorage.com/cosmos/
@audivit’s generation loss module from orhack: https://patchstorage.com/orhack/
@baptiste’s page management system:
https://patchstorage.com/o-knob/
@samesimilar’s gen external wrapper https://github.com/samesimilar/gen_ext/tree/master
Compressor design by Varun Nair: https://designingsound.org/2013/06/28/tutorial-a-compressor-in-pure-data/
Graham Wakefield and Gregory Taylors GSOT book for thoughts on delay implementation in general.
Let me know if I’m missing anyone else!