Stochastic music generation
Hear your ideas
before you write them.
SONATUM is a stochastic music generation engine for contemporary composers. Define a process. Generate material. Decide what's worth writing.
The problem
The cost of testing ideas
Contemporary composition is slow — not because of notation, technique, or lack of ideas, but because ideas are expensive to test.
Most compositional decisions only reveal their validity in sound and in context: whether a process holds over time, whether density should increase or thin out, whether tension accumulates or collapses too early, whether a textural idea scales or breaks.
The only way to find out has always been to write it — which takes hours. And if the idea doesn't work, those hours are gone.
SONATUM exists to collapse that delay.
What it is
A decision surface,
not a composer.
SONATUM generates art music using explicit theoretical models — Hindemith, Palestrina, Schoenberg, Persichetti. No corpus training. No style imitation. No cloud inference.
Every output is the direct consequence of your parameters and its models — not of prediction or learned behavior. All generation runs locally, offline, on your machine.
The moment you start worrying about engraving,
SONATUM has already done its job.
What it is
- A prototyping engine for musical ideas
- A tool for working at the level of intention
- A way to hear whether a process sustains interest
- Local, offline, no cloud required
What it isn't
- A finished-piece generator
- A style imitator or AI composition tool
- A score editor or DAW replacement
- Corpus-trained or cloud-dependent
Workflow
From intention to sound
in seconds.
-
01
Start from structure
Load a template with predefined instrumentation and a generator ready to run — or define your own sections and parts.
-
02
Configure your process
Set parameters that shape musical behavior: register, density, consonance tendency, voice count, rhythmic character. You are biasing judgment, not programming notes.
-
03
Generate
Click Generate. SONATUM writes material into the section locally, in seconds. Each run is a different valid realization of the same process.
-
04
Listen critically
Does the texture sustain interest? Does density feel right? Listen for behavior, not details.
-
05
Iterate or retouch
Regenerate with adjusted parameters, or use the built-in editor to fix local details. Edit sparingly. Regenerate boldly.
-
06
Export when ready
Once a direction is confirmed, export to MIDI or ABC Notation and continue in your DAW or notation environment. SONATUM's job ends there.
Generators
Three generators.
One principle.
Each generator approaches musical material differently. All work from the same idea: you define constraints and tendencies — the engine renders plausible realizations.
Primary
Multiline Generator
The general-purpose generator. Produces everything from a single line to dense polyphony — loose counterpoint to compact homophonic blocks. Key parameters support envelope animation for time-evolving behavior.
Best for: textural exploration, formal trajectories, evolving polyphony.
Legacy
Atonal Line Generator
A focused single-voice generator for freely atonal melodic material. Fine-grained control over contour, interval behavior, rhythmic profile, and directional balance.
Best for: solo lines, inner voices, raw melodic material.
Legacy
Atonal Harmony Generator
A homorhythmic generator for chordal, block-texture writing. Produces freely atonal harmonic succession — intentional, controlled, stylistically coherent without implying tonality.
Best for: harmonic sketching, background textures, chord-block writing.
Audience
Built for composers who work
with process and intention.
SONATUM is designed for composers working in experimental, contemporary, and research-driven contexts — those who approach composition through structure, form, and process rather than improvisation or preset loops.
- You use explicit compositional systems or process-based methods
- You work with art music, new music, or experimental forms
- You spend significant time testing ideas that are expensive to verify in notation
- You value speed in the ideation phase
- You want a tool that integrates with your existing workflow, not one that replaces it
If you work with structure, process, and intention —
SONATUM will fit your workflow immediately.
Features
What's inside.
Three stochastic generators
Multiline, Atonal Line, Atonal Harmony — each with distinct scope and control surface.
No training, no cloud
All generation happens locally, offline. Output is determined entirely by explicit theoretical models and your parameters.
Animated parameters
Envelope-driven control over key parameters across the section's duration. Formal gestures, not automation.
Built-in score editor
Minimal notation interface for retouching generated material. Sufficient for intervention; not meant for engraving.
Onboard synthesizer
Audition material immediately without leaving the app or rendering externally.
Batch editing
Select multiple score elements and edit pitch, duration, or voice simultaneously.
MIDI & ABC Notation export
Move to your DAW or notation software when the material has proved itself.
Template system
Load preset instrumentations with generators pre-bound and configured. Open → generate → listen.
Transposition & pitch scaling
Structural transforms for registral exploration, doublings, and interval augmentation/compression.
Privacy
Not a tracker.
SONATUM does not collect, maintain, or transmit any representation of the user who installs or uses it. All generation, playback, and editing happens entirely on your machine. No telemetry. No analytics. No network calls.
SONATUM is distributed via MuseHub, which operates under its own privacy policies. For information on how MuseHub handles user data, see their privacy policy.
Get started
Free. Local. Offline.
Available on MuseHub — no subscription required. Runs entirely on your machine, with no network dependency.