Papers GAN Music Generation

[marja-w]

• Musegan: Multi-track sequential generative adversarial networks for symbolic music generation and accompaniment: human-AI cooperative music generation, repository: https://salu133445.github.io/musegan/

[marja-w]

Polyphonic music generation generative adversarial network with Markov decision process: https://link.springer.com/article/10.1007/s11042-022-12925-w

• use LSTM network as generator
• the generated sequences are considered incomplete, and completed with a Monte-Carlo Tree search (MCTS)
• the completed music sequence is forwarded to the discriminator, which is a CNN
• CNN outputs reward, and generator is updated using the policy gradient

We could replace the MCTS with our Discrete Event Simulator