M4L.RhythmVAE: Variational Autoencoderを用いたリズム生成DAWプラグイン — AIを用いた音楽制作の民主化に向けて

2023 RESEARCH AI MUSIC

Variational Autoencoder(VAE)-based rhythm generation device for Ableton Live/Max for Live. This plug-in software allows artists/musicians to train and use the rhythm generation model within the music production software(DAW). 

Users of the device can drag and drop MIDI files containing rhythm track and train the model. Once trained, they can automate the transition of the latent vector of the VAE model and generate dynamically changing various rhythm patterns.

The device simplifies the otherwise complicated and cumbersome process of training and utilizing generational models and allows artists to explore the possibility of AI freely. We aimed to democratize the use of AI in the music production process.

Download:

https://github.com/naotokui/RhythmVAE_M4L/releases/

Source Code:

https://github.com/naotokui/RhythmVAE_M4L/

Publication:

Towards democratizing music production with AI-Design of Variational Autoencoder-based Rhythm Generator as a DAW plugin https://arxiv.org/abs/2004.01525

Overview
Training Process
Generation Process

Participants of MUTEK AI Music Lab Tokyo used M4L.RhythmVAE to develop their own musical ideas