Building a system that generates poetry based on real-time environmental data. Temperature, humidity, and ambient sound levels influence word choice, rhythm, and structure.

The goal is to create a living poem that evolves with its environment, blurring the line between observer and observed.

Exploring the intersection of neural networks and sound synthesis. Using machine learning to generate unique timbres and textures that would be impossible to create with traditional synthesis methods.

Initial findings suggest that training on field recordings produces more organic results than synthetic training data.

Investigating complex feedback loops in modular synthesis. Using probability and chaos theory to create self-evolving patches that drift and mutate over time.

Early tests show promising results with Markov chains controlling patch routing. The system becomes a collaborator rather than an instrument.