zhaojw1998 / DAT-CVAE

Codes and MIDI demos of ISMIR 2022 paper: Domain Adversarial Training on Conditional Variational Auto-Encoder for Controllable Music Generation
MIT License
19 stars 1 forks source link

Domain Adversarial Training on Conditional VAE.

Paper: Domain Adversarial Training on Conditional Variational Auto-Encoder for Controllable Music Generation

This is a follow-up work of EC2-VAE, aiming at better conditional control.

Training and validation data is now available here (~2GB). This file contains ~2K processed lead sheets (melody + chord) collected from Nottingham and POP909.

Demo page at https://zhaojw1998.github.io/DAT_CVAE.

Contact

Jingwei Zhao (PhD student in Data Science at NUS)

jzhao@u.nus.edu

Aug. 26, 2022