The aim of this project is to understand to what extent Artificial Intelligence can augment human performance in music production and music criticism. 
The rework of the tracks is generated by Dance Diffusion, a family of audio-generating machine-learning models. We used glitch-440k and jmann-large-580k. The text is generated by GPT-3, an autoregressive language model that uses deep learning to produce human-like text. The images are generated by DALL-E and Stable Diffusion, deep learning text-to-image models. 
We are trained on the following dataset of files: Coil, SPK, Men's Recovery Project, Harsh Noise, Black Metal, Sun City Girls, Cormac McCarthy, Ancient Methods, Italo Disco, Emil Cioran, Bruce Sterling, Nick Herbert, Dino Campana, Laibach, Andrzej Zulawski.
email
metastableai at gmail dot com​​​​​​​
we also make random playlists
↑Back to Top