I1) Why I created a bridge between two worlds

Where does this article fit in?

This article opens the Introduction (I) series. It presents the genesis of the BP2SC project (Bol Processor to SuperCollider, a transpiler from BP3 grammars to SuperCollider code) and links the presentations of I2 and I3.

The Trigger

It was by chance that I met Bernard Bel while I was at a Yukido workshop with Andréine Bel, his wife (in connection with my therapeutic support activities). We quickly became friends, sharing a strange passion for musical formalisms, Indian classical music, and musical modes. It was therefore quite natural that he introduced me to the Bol Processor.

Here was a system created 40 years ago that elegantly solved problems I had only superficially explored: that of formal grammar (a set of rules describing how to construct valid expressions in a language) enabling complex musical expression. This led to the formalization of concepts such as polymetry (superposition of different metric structures), repeated structures with variations, and finally, grammars as a compositional tool.

But when I wanted to use it… frustration.

BP3 (Bol Processor 3, the current version of the software) generates MIDI (Musical Instrument Digital Interface, a standard communication protocol between digital instruments). Period. In 2024, I wanted live coding (real-time musical programming in front of an audience), real-time synthesis, integration with my tools. MIDI is fine, but it’s a 1983 format.

The gap between BP3’s conceptual power and its practical output seemed too limited to me.

My Journey (Short Version)

I am not a musicologist — well, an enlightened amateur. In reality, I could claim different labels, and above all that of an enthusiast: passionate about music, systems, complexity, humanity, and computer science.

Master’s degree (DEA) in networks in the 2000s. On the coding side, never an official career — but twenty years of passionate amateur practice: home automation, Arduino, the warez scene and demomakers in the 90s-2000s. The kind of background that doesn’t fit on a CV, but that forges a true technical culture.

And always, music. Synthesizers, production, experiments. Without formal training, out of passion and curiosity.

These two worlds remained separate for a long time. Until I realized they could come together.

The Observation

BP3: Powerful but Isolated

The Bol Processor is an intellectual gem. Its polymetric notation is elegant. Its formal grammars allow for the expression of structures that traditional scores struggle to represent.

But its ecosystem is restricted:

  • Modest community
  • MIDI output only
  • No integration with modern tools
  • Dispersed documentation

SuperCollider: Flexible but Unstructured

SuperCollider is the opposite. Rich ecosystem, active community, infinite possibilities.

But writing structured music in SuperCollider requires effort. Patterns (SuperCollider’s pattern-based sequencing system) are powerful, but you have to build your own architecture. Nothing guides you towards a coherent organization.

Much SuperCollider code looks like this: bits that work, stacked without an overall vision.

The Missing Bridge

What if we could:

  • Think structure in BP3 (grammars, rules, polymetry)
  • Realize sound in SuperCollider (synthesis, real-time, live coding)

Not just convert MIDI. Truly preserve the structure, allow modification, retain semantics (the musical meaning of constructions, not just the notes).

This is the bridge I decided to build.

The Philosophy

Formalism ≠ Rigidity

There’s a common misunderstanding: to formalize is to constrain. Grammars, rules, specifications — it seems rigid, academic, anti-creative.

I think the opposite.

A grammar doesn’t tell you what to write. It gives you a framework for exploration. Just as the rules of a game enable play, the rules of a grammar enable music.

BP3 doesn’t compose for you. It gives you a language to express your structural ideas. It’s up to you to fill them in.

Structure = Freedom

Apparent paradox: the clearer the structure, the freer you are to manipulate it.

A melody written note by note in a sequencer (music recording and editing software like Ableton or Logic)? Modifying its structure requires starting all over again.

A melody defined as a BP3 grammar compiled to SuperCollider? You change one rule, and the entire instance updates.

Explicit structure is liberating.

Code as a Score

I believe that code can be a musical medium.

Not code as a technical means, hidden behind an interface. Code as a direct expression of musical intent.

When I write:

 

S → INTRO COUPLET REFRAIN COUPLET REFRAIN OUTRO

 

I’m not programming. I’m composing. The code IS the score.

Why Now

The Maturity of Tools

SuperCollider is 25 years old. music21 (a Python library for musical analysis) is robust. Formats like MusicXML (an open XML format for score representation) are stabilized.

The building blocks exist. They just needed to be assembled.

AI and Generation

Everyone is talking about generative AI. DALL-E for images, GPT for text, and now Suno or Udio for music.

But generate what? Statistically plausible sequences of notes? Pieces that “sound like” without understanding why?

I believe that true musical generation comes through structure. Not “here are 100 notes that sound like Bach” but “here is a grammar that captures Bach’s logic”.

BP3 and similar systems are more relevant than ever — not to replace AI, but to give it a structural backbone.

My Position

I’m not in a lab. I don’t have institutional funding. But I have:

  • Twenty years of self-taught technical practice
  • A profile as a musician and systemist, with a familiarity with complexity and transversality
  • A passion for the subject
  • Time to dedicate to it

Sometimes, bridges are built by those who find themselves between two banks, not by those who inhabit one or the other.

What BP2SC Brings

For BP3 Users

Your work is no longer confined to MIDI. Your grammars become executable, modifiable, and integratable SuperCollider code.

You can:

  • Play your compositions with elaborate synthesizers
  • Modify parameters in real-time
  • Integrate BP3 into live coding workflows

For SuperCollider Users

You gain a structured notation system. Instead of building your architecture from scratch, you import a grammar.

The structure is explicit. Patterns are named. Relationships are clear.

For Researchers

A formally specified bridge between two systems. Testable, verifiable, documented.

Not a hack that works for two examples. A systematic translation with 198 passing tests.

What Remains to Be Done

BP2SC v1.0 works. But it’s only one step.

Structure Inference

What if we could go the other way? Take a flat MIDI file and extract a BP3 structure from it?

This is the problem we are currently working on. First results: 83% success on our tests.

The Ecosystem

A tool without a community is a dead tool. We are preparing:

  • Comprehensive documentation
  • Accessible tutorials
  • Varied examples
  • Open source code

What’s Next…

But that’s for later. Some projects take time to mature.

Conclusion

I created this bridge because I needed it. Because I wanted the rigor of BP3 and the flexibility of SuperCollider. Because I believe that formalism and creativity are not opposed.

If you share this vision, welcome. The tools are here. The code will soon be public. The rest we write together.

In This Series

  1. I2 — Bol Processor: 40 Years of Innovation
  2. I3 — SuperCollider: The Swiss Army Knife of Synthesis
  3. Why I created a bridge between two worlds ← you are here

Tutorials :

  • Install BP2SC in 5 minutes
  • Your First BP3 Grammar
  • Analyze MIDI with Structure Inference

roomi-fields explores the intersection between formal musical grammars and modular synthesis.