Bundles ·
Extensions ·
Max Audio ·
Max MIDI ·
Modulators ·
MIDI Tools ·
Sonification ·
Live Racks ·
Free Tools ·






News
 Insights
Signals
Videos
Press
Manuals
FAQ
Support
About

Account
         


Contact · Privacy · Terms

Combinator: 46-Algorithm Clip Combiner for Ableton Live

COMBINATOR

Multi-Algorithmic Sound Mesh

$59
+ applicable taxes
BUY NOW

Combinator is an Ableton Live Extension that merges multiple MIDI and/or Audio clips using a catalogue of 46 algorithms, writing new clips without ever touching the sources. The algorithms span six MIDI families, Merge, Impose, Temporal, Statistical, Harmonic, and Mosh, and five Audio families, Montage, Impose, Granular, Spectral, and Mosh. Every algorithm operates on the same shared framework: a selectable Anchor clip that most algorithms treat as the primary reference, a Length mode that determines output duration, and the ability to Iterate or Evolve multiple variations as simultaneous output clips to Take Lanes or anywhere else in your set. The result ranges from intricate hocketing to hybrid glitchy multi-sample chops to granular textures stitched together from many different samples, or even a spectral hybrid of multiple audio sources driven melodically by a MIDI clip - all in one dialog. An offline-processing sound design powerhouse.


INTRICATE MESHWORK

  • Designates any clip as the Anchor, the primary reference that most algorithms treat as the carrier, base, or pivot, reassignable from a dropdown that takes effect immediately
  • Chooses output length independently of source lengths: match the Anchor, use the Longest or Shortest source, compute the Least Common Multiple for a full phase cycle, or concatenate end to end
  • Cycles sources across cells, slices, or grains using four Order modes: Round-Robin (fixed sequence), Random (Amount scales departure from anchor), Weighted (per-source probability sliders), or Euclidean (Pulses sets how many anchor cells appear in the distributed pattern)
  • Generates independent stochastic variations with Iterate, or chains each output back as the next generation's anchor with Evolve, which works even on deterministic algorithms since each generation receives a different input
  • Writes results to Session slots, Arrangement, or Take Lanes on the anchor track, with audio results limited to Arrange and Lanes since a multi-cell render can't fit a single Session slot

Amount is the one global control that applies across every algorithm, but what it scales is algorithm-specific rather than a universal depth dial. In Intersect and Difference it widens the timing window within which two notes are considered simultaneous. In Union it controls note inclusion probability. In Morph it controls how far the crossfade travels. The tooltip on every algorithm entry states what Amount scales for that particular case. That approach preserves a single familiar control surface across 46 algorithms while letting each one use it meaningfully rather than mapping it to a generic "intensity" that might be musically irrelevant.

22 MIDI Algorithms / 6 Families
The Merge family covers set operations: Union stacks every clip together, Intersect keeps only notes that coincide, Difference keeps notes that don't overlap, Ghost demotes non-anchor clips to a quieter probabilistic layer, and Sieve interlocks clips by distributing every Nth note to each source. The Impose family takes properties from one clip and applies them to another: Matrix lets you route any of seven note properties from any source independently, Graft keeps one clip's rhythm and replaces its pitches using another's (with Nearest, Absolute, and Interval assignment modes), Transplant copies velocity, probability, or deviation from a donor, and Convolve stamps a whole phrase at every onset of a trigger clip.

The Temporal family rearranges and crossfades across time: Hocket alternates clips cell by cell on a fixed grid or at every onset, Zipper weaves clips in a strict order with repeating, ratcheting, and pitch-compounding, Mask/Gate keeps anchor notes only where another clip is playing or silent, Morph crossfades between clips across the output duration, Phase Drift writes copies sliding progressively out of alignment, and Concatenate lays clips end to end with overlap, crossfade, and continuation passes. The Statistical family learns from and votes across content: Markov builds new material from a transition model trained on all sources, and Vote keeps only the notes most clips agree on at each grid position.

The Harmonic family works with pitch and voicing: Harmonic Stack forces every clip onto a shared rhythmic grid and builds a chord from all sounding voices at each position, and Chord Trace keeps the anchor's rhythm and fills in harmony from whatever the other clips are playing nearby. The MIDI Mosh family applies datamosh-inspired operations to note data: Delta moves the anchor's notes using the pitch and timing changes found in another clip, resetting accumulated drift at a configurable Keyframe interval; Bloom echoes every note the clips disagree on, with decay and falloff written as Live note probability so the layer shimmers differently on every playback; and Weave combines the raw bit patterns of note values using XOR, AND, OR, Rotate, or Byte Swap, with controllable bit depth and wrap behaviour.

NEW IN VERSION 1.1.0: Four simultaneous algorithm slots available in both audio and MIDI modes for multi-algorithmic processing.
24 Audio Algorithms / 5 Families

The Audio Montage family rearranges audio without DSP rendering, using source files directly: Hocket switches between clips on a fixed cell or amplitude threshold, Zipper weaves clips with ratcheting and pitch compounding, Phase Drift slides copies out of alignment, Repitch Mosh transposes and rearranges cells with resampling rather than time-stretching, Retrograde reverses cells within scope regions of configurable size, and Concatenate lays clips end to end with crossfades.

The Audio Impose family transfers properties across clips: Envelope keeps one clip's sound and gives it another's loudness shape using a level follower, Formant transfers the spectral shape of one clip onto another via FFT analysis with bandwidth and tilt controls, and Convolve plays one clip through another as a true convolution impulse response with adjustable length and tilt.

The Granular family rebuilds results from short overlapping grains: Intersplice draws grains from across all clips and can use a MIDI score to transpose each grain independently, and Grain Sort chops sources into grains, sorts them by loudness, brightness, or noisiness, and reorders the output accordingly.

The Spectral family shares a single hand-rolled radix-2 FFT engine and covers eight algorithms: Cross-Synthesis filters the anchor through the other clips' frequency content, Winner Takes All keeps the strongest or weakest clip at each frequency bin, Slit-Scan reads different frequencies from different points in time, Magnitude Swap exchanges magnitude and phase components between clips, Freeze Stamp holds a spectral frame from one point in the anchor and gates it against another clip's amplitude or a MIDI score, Spectral Morph crossfades spectral content over time with magnitude-only or magnitude-plus-phase interpolation, Composite blends all clips spectrally using mean, median, geometric, or harmonic statistics, and Scanline assigns runs of FFT analysis frames to different clips in alternating, Euclidean, or random patterns.

The Audio Mosh family applies corruption and modulation: Delta accumulates block-to-block changes from one clip and pushes the anchor with them, Bits combines raw sample data using bitwise operations at configurable word sizes, DCT Quantize crushes the anchor's detail using another clip's spectral structure to produce JPEG-like blocking artefacts, Cross-Modulation lets one clip drive the anchor's amplitude, ring multiplication, playback rate, or phase position, and Waveshape turns one clip's waveform into a waveshaping curve the other is read through. 

MIDI x AUDIO

  • Routes MIDI clips to a control score when a mixed MIDI-and-Audio selection is processed, rather than synthesising them or treating them as audio
  • Passes onset timing, velocities, pitches, and density from the score to the audio algorithms that read it: Hocket uses onsets for cell boundaries, Zipper for slice boundaries, Intersplice for per-grain transposition, Grain Sort for depth into the sorted grain pool, Freeze Stamp for gating and pitch, Spectral Morph for morph timing, Scanline for frame switching, Repitch Mosh for per-cell transposition, and Retrograde for cell reversal decisions
  • Draws MIDI score ticks over the visualiser canvas only for algorithms that actually read the score, leaving the canvas clean for algorithms that don't

The MIDI-as-score model is the feature that makes Combinator something other than a clip renderer. A MIDI clip doesn't get synthesised into sound in this path, it becomes a set of instructions: when to switch, how loud, what pitch to transpose to, when the gate opens. That means a rhythm clip written for a completely different purpose can score a spectral or granular audio result, and the MIDI clip's own sound is irrelevant to the outcome.
Tool Tip Annotated
Learn all you need to know directly in Live - just click the [?] icon in the Extension and hover your mouse over any parameter.
Scale Aware
Natively integrated with Live 12's scale and key for seamless workflow.
Already comes included in:
Extensions Bundle 001 · Ultra Bundle

Requires:
Ableton Live Suite 12.4.5 or above.
Not available in Live Standard, Intro, or Lite.

EXTENSION
Melodic · Rhythmic · Audio · Transformation · Generator