Skip to content

Latest commit

 

History

History
13 lines (9 loc) · 901 Bytes

README.md

File metadata and controls

13 lines (9 loc) · 901 Bytes

Study No. 1 for AI Ensemble (2021)

An excerpt from the composition

I trained a bunch of AI instruments/performers based on the DDSP architecture by the Magenta team at Google, then composed for an ensemble of those AIs. The composition is constructed in December.ipynb notebook. Graphical score was created using numpy. Final render of the score follower video is on YouTube.

I am currently working on a real-time, 48kHz, Pytorch implementation of the model here. You can see the training/prototyping progress at wandb.