In this mini-experiment, I further explore whether playable “instruments” can be extracted from a single audio recording by overfitting a simple model of resonating objects and a sparse control signal.

Read more: