audiobriefs
10:15in productionCh. 1 · Rhythmic chaos, controlled/ 10:15 · ceiling 15 min
Synths & drum machines · Sound design & sampling

Mutable Instruments Marbles

It doesn’t make music—it makes decisions for you, then learns from them.

Mutable Instruments Marbles is a Eurorack module for generative timing and control voltage. It uses hardware randomness, memory-based reuse (DEJA VU), and adaptive quantization (STEPS) to produce rhythm and pitch that respond to player input. It has no audio output, no presets, no MIDI, and no visual interface beyond LEDs. Its strength is in linking decision-making across domains—timing, pitch, and memory—in real time.

Chapters & takeaways4
  1. 1:35
    Rhythmic chaos, controlled

    The t section creates three interlocked, jittered gate streams from one master clock.

  2. 3:21
    Two kinds of randomness

    X outputs three independent random CVs; Y delivers one slow, smooth, full-range voltage untouched by memory.

  3. 5:19
    Randomness with memory

    DEJA VU chooses whether to repeat or refresh—making memory part of the random process.

  4. 6:36
    Scales you grow, not pick

    STEPS learns scales from your input; six C-rooted slots hold what you teach it.

Worth your time?

Yes. Use the whole thing.

4.5/ 5
What works
  • links timing, pitch, and memory in one module better than any other Eurorack device
  • DEJA VU section uniquely ties reuse probability to live performance
  • STEPS quantizer learns from your input instead of imposing fixed scales
  • t section delivers three interlocking, jittered gate streams from a single source
What does not
  • generate audio
  • store user sequences beyond six scale slots
  • offer MIDI or USB
  • provide patch memory or visual feedback beyond LEDs
Use it if
  • modular players who treat randomness as a collaborator
  • those who want CV that adapts, remembers, and misbehaves in musically useful ways
Skip it if
  • beginners seeking immediate melodic results
  • users needing recall, sync, or audio output
The written brief2 min read

What it is and what it was built for

A Eurorack module for generating structured, responsive randomness in timing and control voltage. Built for live modulation, generative composition, and breaking out of repetitive patterns—not for sound generation or sequencing.

How it works, in terms a player would use

It generates jittered clocks and random CVs in three parallel streams. The t section splits a master clock into three gate outputs. X 1–3 each output independent random voltages, clocked either by individual t outputs or a shared external clock. Y delivers a slow, smooth, full-range voltage unaffected by DEJA VU. DEJA VU toggles between reusing past values and sampling fresh hardware randomness. STEPS quantizes incoming CV by progressively eliminating notes from a chromatic scale based on occurrence frequency. Six C-rooted scale memories are selected via LED feedback and button presses.

What it does better than anything else

It links timing, pitch, and memory in one module better than any other Eurorack device. Its DEJA VU section uniquely ties reuse probability to live performance—past choices directly shape future ones. Its STEPS quantizer learns from your input instead of imposing fixed scales. Its t section delivers three interlocking, jittered gate streams from a single source—no routing required.

What it does badly

It does not generate audio. It cannot store or recall user sequences beyond six scale slots. It offers no MIDI, no USB, no patch memory, no visual feedback beyond LEDs. Its randomness is not tunable in real time without holding buttons and adjusting parameters mid-flow. It gives no indication of internal state beyond blinking LED colour and rate.

What people made with it

People used it to drive drum machines, modulate filter cutoffs, trigger melodic phrases, and create evolving rhythmic textures. Its ability to learn scales from incoming CV made it central to live-keyboard improvisation and modular jamming where tonal coherence emerged from interaction, not pre-programming.

Who should use it, and who should not

Modular players who treat randomness as a collaborator—not a substitute for intention. Those who want CV that adapts, remembers, and misbehaves in musically useful ways. Not for beginners seeking immediate melodic results, or for users needing recall, sync, or audio output.

Is it worth your time

Yes—if you need generative timing and CV that responds to your playing, not just random noise. It demands hands-on engagement: no presets, no menus, no screen. Its value lies in how it interrupts habit and forces new decisions—not in stability or recall.

Same bench · Synths & drum machines4 of 52
8:19
Mutable Instruments BraidsMutable InstrumentsBraids delivers radical timbral variety in minimal space—but with no internal editing, no tuning stability guarantees, and no analogue circuitry. It is a selector, not a synthesizer.
8:40
Mutable Instruments CloudsMutable InstrumentsClouds is a granular audio processor for Eurorack that transforms live input into dense, overlapping grains. It records continuously into a short, variable buffer (up to 8 seconds), allows freezing that buffer to isolate a sonic space, and provides hands-on control over grain position, size, pitch, density, timing distribution, and envelope shape. A diffuser blurs transients using all-pass filters. It supports 40–60 concurrent grains and operates in stereo with extensive CV control, dry/wet blending, feedback, and reverb-like effects during freeze mode. It does not store or recall samples, nor does it support offline processing or waveform editing.
10:15
Doepfer Dark EnergyDoepferA compact, monophonic, fully analog synthesizer released in 2010 as a stand-alone version of the Doepfer A-111-5 Eurorack module. It features dual LFOs capable of audio-rate FM (~5 kHz), a 24 dB/octave low-pass VCF, and a triangle-based VCO. The VCF can function as a sine oscillator when the VCO is disengaged and resonance is maxed. All sound generation and modulation are analog; only USB/MIDI contain digital components. It has no polyphony, no patch memory, no velocity or aftertouch, and no audio input. Superseded by the Dark Energy II in 2012.
9:03
Korg DDD-1Korg · 1986The Korg DDD-1 is a 1986 sampling drum machine with 14 velocity-sensitive pads, 12-bit samples, limited user sampling, five dynamic envelopes, and ROM card expansion (up to four cards, eight sounds each). It has 18 presets and six assignable audio outputs.
Up next in Audio

Make Noise Maths

Make Noise · 9:34

MATHS doesn’t make sound — it makes time, slope, and logic behave.

9:34