UCSB student Dariush Derakhshani, Winner of the Mellichamp Mind and Machine Intelligence's 2024 AI/Human Music Creativity Contest

Pulsar Rays 

Listen here: https://mind-machine.ucsb.edu/media/audio/pulsar-rays-2021-2023-dariush-derakhshani

Artist Statement

“Pulsar Rays” merges advanced machine learning techniques with the natural rhythms of the environment. This piece emulates the intricate symphonies of nature, drawing on biophonic, geophonic, and anthropic sounds synthesized through stochastic models. Central to its creation is the IRCAM RAVE model, which combines digital sound synthesis with the organic qualities of natural elements like water, stone, and sand. Inspired by Roche's dissertation on machine learning in sound synthesis, the RAVE model employs Variational Autoencoders (VAEs) to capture the probability distribution of audio data, enabling the generation of diverse and realistic sound variations.

The model's two-stage training process—Representation Learning and Adversarial Fine-Tuning—ensures high-fidelity sound synthesis, crucial for the real-time demands of modern music. Pulsar Rays is a testament to the potential of integrating human creativity with machine learning, recreating the complex dynamics of ecological acoustics to produce an immersive auditory experience. Despite challenges, such as handling low-frequency sounds, the result is a composition that pushes the boundaries of sound design and musical creation.