Toolkit for analyzing just intonation. Where notes relate to each other through simple whole number fractions instead of the societal construct of equal temperament.
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 system is built as a distributed high-signal pipeline:
Anceps is now cloud-native. You can run the heavy inference and rendering layers on Google Cloud while keeping your audio analysis local.
Deployed on Google Cloud Run.
https://anceps-inference-hxaum2omaa-uc.a.run.app/ws.Hosted on Google Cloud Storage.
To connect your local SuperCollider/Octave stack to the cloud:
git clone https://github.com/ejcohe22/anceps.git
cd anceps
docker compose up octave
This will listen for local OSC messages and push prompts to the Cloud Run backend.
sc/main.scd and start the analysis engine.To deploy your own instance to GCP:
chmod +x deploy.sh
./deploy.sh
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.
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.
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.
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.
MIT / Unlicense. This is free and unencumbered software released into the public domain.