dschihadelay by monohusche: Randomised Delay From Apparat's Patch
dschihadelay is a single-tap delay that refuses to stay still. monohusche built it from code in "dschihad", the performance patch used by Apparat and Modeselektor, and the design intent shows: the delay time is synced to Live's transport, but almost every other parameter is being randomised while you play. Delay time, feedback amount, filter cutoff and panning all move on their own, inside ranges you set.
That makes it a different animal from Live's stock Delay. Stock Delay is a tool you dial in and leave. dschihadelay is a tool you aim: you decide how wide the randomness is allowed to roam, and the device picks values inside those walls.
What is actually randomised
Four things move:
- Delay time — synced to the transport, so the taps land on grid divisions rather than drifting into arbitrary milliseconds. The randomness picks which division.
- Feedback, with a freeze option. Freeze is the performance control here: it locks the current feedback state instead of letting the next random value wipe out whatever just built up.
- Filter cutoff, with lowpass, highpass, bandpass and bandstop available. Each repeat can land in a different part of the spectrum.
- Panning. The device processes both channels, so the stereo field moves with the repeats rather than smearing them into a mono blob.
The controls that matter most are the min/max dials. Each randomised parameter has a pair of them, they are synced, and they are automatable in Live. That is the whole workflow: you are not automating the delay time, you are automating how much freedom the delay time has. Narrow the window to almost nothing and dschihadelay behaves like an ordinary synced delay. Open it up and the tail scatters.
Filter placement is the real tone control
The filter can sit inside the feedback loop, after the delay unit, or in both places at once. These are three genuinely different results and it is worth auditioning all three on the same source before you commit.
Inside the loop, the filtering compounds: each pass through the feedback path gets filtered again, so repeats degrade cumulatively and tails die off in a shaped way. With a randomised cutoff that means each successive repeat can be darker or thinner than the last in an unpredictable order, which is where the device earns its character.
After the delay unit, the filter treats the whole wet output once. Repeats keep their body; the filter becomes a colour sweep over the tail rather than a destruction process. This is the setting to use when you want the delay to stay legible against a busy arrangement.
Both at once is the aggressive option, and combined with bandpass or bandstop it will do things to a feedback tail that are hard to reproduce any other way.
Three ways to drop it into a session
1. Wet-only on a return track. dschihadelay can output either a wet/dry blend or wet only, and the wet-only mode exists precisely for return duty. Put it on Return A, set the min/max windows fairly narrow to start, and use the send from a drum bus or a single percussion channel. Automate the send rather than the device. Then, once you know what the ranges sound like, automate the max dials open across an eight-bar fill. The track stays dry and intact; the return track falls apart on cue.
2. Insert on a vocal or lead, with feedback freeze on a key. Here you want the wet/dry output so the dry signal stays in place, and the filter placed after the delay so the repeats remain identifiable as the vocal. Map feedback freeze to a controller button. Throw a word into the delay, hit freeze, and the tail holds while the next phrase comes in clean over the top. Because panning is randomised, the frozen tail spreads out instead of stacking in the centre where the dry vocal lives.
3. On a looped synth or noise source, as a generator. Feed it something sustained and harmonically simple, open the delay time range wide, put the filter in both positions, and record the output to a new audio track for several minutes. You are not mixing at this point, you are harvesting. Chop the result into one-shots and transitions. Since the delay times are transport-synced, the material you capture stays usable at your project tempo rather than needing time-stretching.
A practical note for all three: randomness plus feedback is a fast route to an uncontrolled level spike. Put a limiter after it on the return, and watch the output on something like jo.Floating Oscilloscope, which can run in a floating window so you can keep an eye on the waveform while you are working elsewhere in the set.
What the numbers tell you
29,886 downloads and a 4/5 rating across 14 comments, on a device last touched on 4 January 2010 and marked version 1.0. That combination says something specific: this is a device people kept finding and kept using long after its author stopped working on it. A 4/5 rather than a 5/5 is consistent with a device that is strongly opinionated — randomised delay is not for every session, and some people will have downloaded it, found it too wild for their work, and said so.
The flip side of the age is compatibility. It targets Live 8.1 and there is no record of an update since early 2010. If you are running a current Live, treat the first load as a test rather than a commitment: drop it on a scratch track, confirm it passes audio and that the dials respond, and only then build a part around it.
Where it sits among comparable audio effects
Within the same download bracket in the audio effect catalogue, the closest sibling in spirit is College Dropout by CaligulaCuddles — also built on controlled randomness, but aimed at tape and vinyl failure: random volume dropouts and pitch warping. It is the newer and better-maintained of the two, at version 2.2, last updated December 2019, targeting Live 10.0.5 and Max 8.0.2, with a 5/5 rating across 30,944 downloads. If randomness is what you want and modern compatibility is non-negotiable, start there; we covered it at length in the College Dropout deep dive. dschihadelay is still the one to reach for when you want the randomness applied to a delay tail specifically.
Pitch Drop by bitbutter is from the same era — March 2010, Live 8.1, 4/5, 35,999 downloads — and is the opposite design philosophy: one effect, one job, turntable spin-down at an adjustable rate, no clicks. Pair them if you like. Pitch Drop ends a section; dschihadelay fills the hole afterwards.
3X Parameter Sequencer by lvehon approaches movement from the control side instead. It uses the Live API to sequence parameters on Live's own devices and on plug-ins, so it is the deterministic answer to the same problem: where dschihadelay rolls dice inside your ranges, 3X plays a written sequence. At 23,749 downloads and 5/5 with 23 comments, and last updated in 2012, it is the choice if you need the same move to happen the same way every time.
Who should skip it
Skip dschihadelay if you need recall. A randomised delay will not reproduce last night's take, and if your workflow depends on bouncing stems that match a reference mix exactly, this device will fight you unless you print the audio immediately.
Skip it if you want a clean dub delay. Transport-synced repeats with a stable filter and predictable feedback are a solved problem in Live's stock devices, and you do not need randomness to get there.
And skip it if you are unwilling to troubleshoot a 2010-era device on a modern Live install. The maintained randomness tools in the current audio effect roundup are a safer starting point if you are building a set you have to trust on stage.
Everyone else — anyone who performs with a laptop, builds transitions by hand, or wants a delay that generates ideas rather than repeating them — should load it, open the min/max windows wider than feels sensible, and record whatever comes out.