Keisuke Nohara, Ryosuke Nakajima

Algo-Rhythm (2019). This work is a live drumming performance between human and AI. AI interprets the rhythm played by the human drummer and generates unique rhythm patterns in real-time, using GrooVAE model based on the Recurrent Variational Autoencoder (VAE) architecture in the MusicVAE framework of magenta. Human player also responses to AI’s rhythm and tries to change his own performance and play new rhythms against AI. As the autonomous system and a human interact with each other, the leading entity becomes indistinguishable, creating novel rhythmic sensations that can only be possible with this AI-human collaboration. In constantly changing performance flow, visualization shows high dimensional latent vectors of AI transformed into a 3D space, rhythm patterns played by both players and other musical information to represent the dialogue between AI and human to the audience. This project explores the potential for new musical expression in the AI-human relationship, aiming to deviate from a fixed and formalized rhythmic structure.