Tangent
FR
Back to projects
•
2 min read
JMON labs

The Problem

Algorithmic music composition typically requires complex environments, specialized languages (SuperCollider, Max/MSP), or heavy libraries that are difficult to learn and share. Generative approaches — minimalism, stochastic processes, cellular automata — combine music theory and algorithmic flexibility, but remain poorly accessible. The Python package Djalgo was developed at tangent.to to fill this gap. However, Djalgo lacks a clear, flexible, comprehensive, and declarative structure for defining musical items. Moreover, Djalgo’s Python foundation remains too heavy for installation-free execution via WebAssembly.

The Approach

JMON (JSON Music Object Notation, jam on!) is a rich and declarative format for describing music. It enriches the MIDI format and makes it readable in a text file.

JMON-algo is a JavaScript framework for algorithmic music composition based on the JMON format. It provides:

  • A structured, readable, and programmable JSON music representation
  • Integrated music theory tools (scales, progressions, harmony, rhythm)
  • Generative composition techniques without deep learning: random walks, fractals, cellular automata, genetic algorithms, Gaussian processes
  • Seamless conversion to multiple formats (MIDI, ABC, SuperCollider, Tone.js)
  • Interactive score visualization and playback

The framework favors mathematical and algorithmic creativity over artificial intelligence, offering precise control, clear understanding of compositional processes, and ownership of the creative process.

Technical Implementation

  • Pure JavaScript framework (converted from TypeScript for better compatibility)
  • Integration with Observable for interactive notebooks
  • Specialized modules: format validation, MIDI/ABC/SuperCollider conversion, score visualization
  • Musical analysis tools (useful for genetic algorithms)
  • Usage via npm or directly in Observable

Current Status

The project is active, but for a complete and functional approach, it is currently recommended to use Djalgo. A free interactive book is being written. Future developments are planned to fix bugs and expand analytical capabilities and composition tools.