UC San Diego

Our system generates a class of musical works. It does so without human intervention by using Lorenz attractors at multiple time scales to construct several time series for controlling sound synthesis parameters. It is implemented with Python, SuperCollider, and Open Sound Control. An indefinite number of works can be autonomously generated, yet each is unique: each exhibits unique musical form globally and unique sonic transformations locally. We discuss the system in the context of (1) Laske's composition theory and (2) phenomenological perspectives on chaos and self-similarity.