Co-Explorer

Hugo Scurto et al., 2019
Max/MSP, Python, deep reinforcement learning

An interactive AI that support sound exploration through positive or negative feedback. It uses deep reinforcement learning algorithm to autonomously explore sounds while interactively learning a person’s preferences. It supports a variety of human-AI interactions designed to balance agencies along exploration. The Co-Explorer lies at the heart of the ægo performance, and was applied to visuals in other artworks (see Trëma by V. Drouin & D. Perreault).

Designed collaboratively with sound artists.
Year
2019
Credits
The project was developed with Bavo Van Kerrebroeck, Baptiste Caramiaux, and Frédéric Bevilacqua in collaboration with the ISMM group of IRCAM and the ex)situ team of LRI (INRIA), in the context of a PhD thesis at Sorbonne Université.
Publication
Article at ACM TOCHI (2021)
Events
Exhibition @ HMC 2022 (Dec.02.2022)
Workshops @ Ircam (Jul.19.2018)
Exhibition @ ISMIR 2018 (Sep.24.2018)
Code
GitHub

This slideshow requires JavaScript.