Audio Forager

Audio Forager is an experiment that uses machine learning to arrange more than 44,000 sounds from Freesound in a 3D map. You explore the sound space, select individual sounds to keep and play them again.

Explore and make instruments

You can use the selected sounds to build a drum set or other instruments, such as a piano made from synth notes or a marimba made from pops and crackles. You can also explore the sound data without making or playing an instrument.

The sound data comes from a dataset created by Kyle McDonald, who scraped the audio and made t-SNE embeddings of it. The project was created during Unity Artist in Residency 2017.