We’ve been working on some tech groundwork, and making a prototype synth. It uses machine learning to map from a 3D parameter space (a 3D joystick) into a bigger parameter space of an FM synth. The prototype has controls for collecting training data, and for training the model. In the background, it uses an XMOS chip to run the neural network and the sound synthesis.
We’re happy to announce that we’ll be running a PhD Course on Musically Embodied Machine Learning at AAU Copenhagen, in collaboration with Dan Overholt. The...
Andrea presented the project at the Music RAI workshop at QMUL, on ethical and responsible music making, and bias in AI music generation.
We presented the project at the Sussex AI day 2024.
Before the project officially began, Chris presented a prototype at the Sussex AI launch event. It was a DiffLogic network running on an RP2040 microcontro...
