From Generality to Mastery: Composer-Style Conditioned Music Generation

• Mingyang Yao and Ke Chen "From Generality to Mastery: Composer-Style Symbolic Music Generation via Large-Scale Pre-training."
  Conference of AI Music Creativity (AIMC), 2025

Model Architecture

"Generality" Stage

• Model backbone: 12-layer Transformer with relative positional encoding
• Num trainable params: 39.6M

"Mastery" Stage

• Model backbone: 12-layer Transformer with relative positional encoding plus adapter modules inserted after every two transformer layers
• Num trainable params: 46M

Citation

@inproceedings{generalitymastery2025,
  author = {Mingyang Yao and Ke Chen},
  title = {From Generality to Mastery: Composer-Style Symbolic Music Generation via Large-Scale Pre-training},
  booktitle={Proceedings of the AI Music Creativity, {AIMC}},
  year = {2025}
}