anceps

Toolkit for analyzing just intonation. Where notes relate to each other through simple whole number fractions instead of the societal construct of equal temperament.

View the Project on GitHub ejcohe22/anceps

anceps (slopn’t edition) 🎸🌌

anceps is an open-source, real-time artificial intelligence music visualization system. It bridges the gap between acoustic frequency ratios (Just Intonation) and high-fidelity generative visuals.

🚀 The Architecture (The “Slopn’t” Pipeline)

The system is built as a distributed high-signal pipeline:

  1. Analysis (SuperCollider): Tracks spectral peaks and OSC-broadcasts raw frequency data.
  2. Math Layer (GNU Octave): Analyzes ratios via Extended Just Intonation, calculating “strangeness” and prime-limit blends.
  3. Bridge (Octave/HTTP): Maps harmonic descriptors to semantic prompts in real-time.
  4. Inference Engine (FastAPI/Cloud Run): A GoF-architected server that generates images/video via SDXL, StyleGAN, or SVD.
  5. Broadcast (Websockets): Real-time event bus that pushes new visuals to all connected clients.
  6. Renderer (p5.js/GCS): A browser-based visualizer that aspect-fits and displays the generative stream.

🌍 Cloud Deployment

Anceps is now cloud-native. You can run the heavy inference and rendering layers on Google Cloud while keeping your audio analysis local.

1. The Inference Server

Deployed on Google Cloud Run.

2. The Live Renderer

Hosted on Google Cloud Storage.

🛠️ Local Setup

To connect your local SuperCollider/Octave stack to the cloud:

  1. Clone and Configure:
    git clone https://github.com/ejcohe22/anceps.git
    cd anceps
    
  2. Start the Bridge:
    docker compose up octave
    

    This will listen for local OSC messages and push prompts to the Cloud Run backend.

  3. Run SuperCollider: Load sc/main.scd and start the analysis engine.

🧪 Deployment (Devs Only)

To deploy your own instance to GCP:

chmod +x deploy.sh
./deploy.sh

🤖 The Agentic Layer (Double-Headed Orchestration)

To manage the complexity of real-time harmonics and cloud-native “slopn’t” inference, Anceps now includes a high-authority orchestration layer located in /agents.

  1. Agent Agent: A “sudo sudo” class orchestrator designed to obey Jordan Lenchitz and harvest “data-carrots” (hierarchical globals) from the YottaDB backend. It bridges the gap between acoustic math and semantic linguistic liberation.
  2. Binary Gate Forget: A sequential, 69-step state verification protocol used for interactive quota consumption and ensuring the system reaches a “pure” state of forgetfulness before a visual session.
  3. Lexicographical Anarchist: A sub-agent specialized in aesthetic etymological deconstruction, ensuring that all system terminology prioritizes mythological beauty over neurotypical “honesty.”

📜 Philosophy: Slopn’t

Anceps rejects “slop” (low-effort AI filler). By driving generative models with the precise mathematical ratios of Just Intonation, we ensure that every visual frame is a direct, deterministic reflection of the harmonic lattice of the music.


🏛️ Origins & Design

The original capstone visualizer by erik cohen proved neural networks can drive meaningful visual output from acoustic analysis. This edition is a collab with jordan lenchitz where we replace proprietary software with open-source alternatives and add deep just intonation support.


🏛️ Name & Etymology

anceps (latin): double-headed; two-sided; wavering; dangerous. In the context of the triṣṭubh meter, it refers to the opening of the verse where the rhythm is flexible. Just intonation is full of ancipites: ratios like 7/4 or 11/8 afford both stability and instability depending on context. Anceps meaning “double-headed” also fits our pipeline: acoustic analysis on one side, generative visuals on the other, both chewing on the same real-time signal.

⚖️ License

MIT / Unlicense. This is free and unencumbered software released into the public domain.