Recombinator allows you to take a buffer, mangle it into an absolute mess, and then — presto! Abracadabra! — restore it to its original beauty. Even more than that, it performs these functions probabilistically (although they can also be controlled at will), creating delays and loops that mutate over time, sometimes changing, sometimes repeating those changes or changing the already changed audio more, sometimes smoothing things back over only for more changes to occur. I think it’s super fun and weird and wonderful. I really like this one, but I also acknowledge that it is a lot.
The basic idea of the patch is that there are three signal paths — Recombine, Replace, and Restore. Recombine and Restore are made from delay lines, good, stable delay lines. Replace is made from alternating read/write loopers, which can have their pitch/speed, playback direction, and start position modulated — very unstable stuff. Each of these signal paths can be fed back into the Recombinant path: when it feeds back into itself, whatever was going on is preserved and repeated; when the Replace signal path is selected, glitches, pitchshifts, and reversals are introduced; and when the Restore signal path is selected, the original, unadulterated signal returns everything to its original state. It’s the play between these signal paths that really makes things special. Glitches appear and repeat and disappear again, or get compounded by being fed further into the loopers, or only appear once. They can crop up infrequently or quite often. There’s a lot of play at work here.
There are a lot of options, so I want to jump into the controls. You might watch the walkthrough video or read these patch notes and get overwhelmed, but this is really a patch meant to be played with — load it up and mess around with controls and see what happens. There are no wrong answers.
The signal path is stereo throughout, although to avoid some irritating stereo effects caused by decoupling the two sides (more on this later) a bit of the wet signal path is fed into both outputs. (Right is mostly right, but a little left and vice versa.) The dry signal path is true stereo.
Stompswitches (all of these behaviors are replicated in buttons on the front page, so if you are using this with Zebu, you can perform all of these functions):
Left — tap tempo; the patch also accepts MIDI clock. There is a tap tempo button on the second row.
Middle — replace override (momentary). This overrides whatever signal switching is determined by probability and makes the replace signal path the active one. This function can also be replicated by pressing the Replace UI button in the top left corner.
Right — restore override (momentary). This overrides whatever signal switching is determined by probability and makes the restore signal path the active one. This function can also be replicated by pressing the Restore UI button in the top left corner.
External — loop. This essentially maxes out the feedback control. There is also a LOOP button located on the second row.
I’m going to move through the controls in corners, for the most part, as these comprise different control paramaeters.
In the top left corner is a UI display which shows which of the three SIGNAL PATHS is active at the moment. As mentioned above, the Replace and Restore UI also function as buttons for overriding the probability controls below.
REPLACE CHANCE — this determines how likely it is that the Replace signal path will be used. Here be glitches.
RESTORE CHANCE — this determines how likely it is that the Restore signal path will be used. Restore always takes precedence over replace, so keep that in mind when dialing in the chances — if restore and replace are both selected, restore will win.
There are also a couple of options located here.
ISOLATE REPLACE — this essentially mutes the recombinant and restore output signal paths. They’re still working in the feedback loop, but we won’t hear them if they are selected by probability; you’ll only hear the Replace path when it is selected. So, what this does, basically, is create audio that sprinkles in different glitches and artifacts and weird things. Less a looper or delay, and more a sort of sporadic accompaniment.
REPLACE TRIGGER — when this button is off, the replace variation (more on this below) is selected each time the clock multiplier controlled by the resolution parameter (more on this below) changes. This means that if, probabilistically, the replace signal path is heard for two or more of these multiplier cycles in a row, the variation may change on each of those cycles. When this button is off, the variation selected will be active until another signal path is selected. It can make things (slightly) less chaotic.
FADE — because the signal paths are selected by switching between different VCAs, and because the amplitude and frequency in the signal paths can be quite different, or they might catch different points in the buffer, there can be some clicks generated by the switching (VCAs proved better than audio switches for this, if you’re interested). The fade control can allow you to smoothly blend between the signal paths, using a fade time determined by the resolution clock. This both makes clicks less pronounced but also changes the general timbre of the patch, making it a bit more mellowed out.
Since I’ve already mentioned them, probably the next place to go is the bottom left corner, where the PROBABILITY MIXER lives.
The probability mixer allows you to select between five different defined variations, depending on the weight you give each variation. You can use fewer variations (I often do, and the default condition of the patch employs only three). More variations means more chaos, but you can also reel things in and keep them simpler by using fewer variations (or just one, by setting all of the others to zero).
There are three controls for each of the five variations:
VARIATION CHANCE — this sets the weight or likelihood of a given variation being selected for the Replace signal path
INTERVAL — or speed-pitch for the variation
REVERSE CHANCE — this determines the likelihood that the loopers play forward (0) or reverse (1), with values in between setting different likelihoods of one or the other outcome
There are also a couple of buttons that allow for options:
SYNC INTERVAL — when this button is off, the interval for each side of the stereo path will be determined independently; they may agree, they may not. It’s a good way to add interesting stereo effects. (Default is on.)
SYNC DIRECTION — when this button is off, the playback direction for each side is determined independently. (Default is on.)
Beside the variation chance controls are a few more controls that affect the replace signal path and variations:
RESOLUTION — this controls the multiplication of the clock used to select between different signal paths (and variations if the replace trigger option is off). From 1x to 32x — the higher the resolution, the glitchier and more chaotic it becomes.
RESOLUTION MOD RANGE — this applies a random value modulation to the resolution control, changing the speed of the resolution clock with each cycle. Basically, chaos central — this is a bipolar control, so the range can be set to make the resolution clock slow down (negative values) or speed up (positive values)
START POSITION RANGE — each clock cycle, the start position of the loopers is recalculated, based on this control which defines the range of a random number applied to the start parameter of the loopers. This is useful if you’re using a variation with the interval set to 0/unity speed-pitch, because it can introduce different parts of the buffer, creating stutters and changing the order of audio
Okay, on to the top right corner:
Here you have a set of radio buttons that determine the BUFFER LENGTH.
Buffer length is always set by tap tempo or MIDI clock, but you have four different options for clock divisions/multiplications:
2X tempo — twice the tempo; shortest buffer
Unity tempo — same buffer length as the tapped tempo
1/2 tempo — double the length of the tapped tempo
1/4 tempo — quadruple the length of the tapped tempo
There are a couple of reasons for these options. The first is that the tap to CV converter struggles with tap lengths above ~6 seconds. Those converters are used in this patch to provide stability to the delay lines. The delay lines can be up to 16 seconds long. So this allows you to reach buffers of that length. But also it makes it easier to tap in longer buffer lengths.
But another benefit is that each time the buffer length is changed, the buffer is reduced in size or doubled in size, and this allows you to play around with capturing shorter loops and doubling or quadrupling them, and allowing different signal path switching to occur across each one. Likewise, you can shorten the buffer and create stuttering effects. There’s a lot of fun to be found in using these buffer lengths as a performance/compositional tool; a miniature time lag accumulator, essentially.
Below the radio buttons are a few controls:
TAP TEMPO — allows you to tap in a tempo (also available via the left stompswitch; MIDI clock will override tap tempo)
LOOP — essentially sets the feedback amount to 1 and continually regenerates the buffer
OVERDUB — when LOOP is active and this control is active, new audio received at the input will be added to the loop. When this is off, new audio will not be added to the loop, if, for instance, you want to play over something captured in the buffer.
ERASE — this erases everything _except_ the contents of the Restore buffer. You can use this to add silences, or to build from scratch again, introducing bits of the restore buffer probabilistically. The fade parameter affects the attack and release of the erase function, so you can soften it a little, if you want to.
Okay, the last section is some pretty straight forward stuff; it’s in the bottom right corner.
FEEDBACK — when loop is off, this sets the decay of the delay (an interesting target for an expression pedal input or CV, since it will loop at a setting of 1)
LOW-PASS FILTER FREQUENCY — you can use this to shape the top end; particularly with pitch up effects, this makes it easier to avoid shrillness that can come with that
DRY LEVEL and WET LEVEL — use these to set the mix. Wet level is just a bit above unity.
The singular change in V2 is relatively simple, but very effective. On the second page (because there was no room on the first), I added a control called REPLACE FEEDBACK. In the original configuration, feedback into the replace loopers was derived from the Recombinated delay line. This meant that changes in speed-pitch and direction compounded: if a sound had already been pitched an octave up, when it fed back into the loopers, if it was once again pitched an octave up, it would be two octaves up. Or it might be pitched an octave down, then a fifth up, resulting in a fourth down. What was reversed might get reversed again, playing forward. And so on. Beautiful, happy accidents occurred, but it was also a road to chaos.
The REPLACE FEEDBACK controls a mix between two sources for feedback.
At 0, the feedback is derived from the Restore delay line. This means that pitch-speed and playback direction effects -will not- compound: instead, new Replace instructions will fully replace what is in the buffer. If audio had been pitched up, and a Replace instruction says pitch down, then the replacement will only be the pitched down sound.
At 1, the feedback is derived (as in the original) from the Recombinate delay line. This means, as in the original, changes to speed-pitch and playback direction will compound, producing fluctuating pitches and playback directions, as different layers of replacement instructions affect one another.
Since it is a mix, you can also have these compounding effects present but mitigated by choosing a value between 0 and 1.
Two things to keep in mind:
Although it was always possible, the additional CPU for this addition makes it likely that adjusting the filter frequency will cause clicking (as it employs a multi-filter). I would think, in most cases, this would be a fairly “set and forget” control, but if you find yourself “playing the filter” a lot, you may prefer V1.
The additional CPU may also affect Euroburo functionality. Again, V1 was largely programmed on Euroburo (outside of adding stompswitches), so it should run fine on Euroburo, if you find V2 does not. Get it contact with me if you would prefer to remove compounding feedback on a V1 iteration of the patch; we can’t add the control, but I could instruct you on how to rewire the internal connections to produce the same result.