The dataset can be downloaded from Zenodo or GitHub.
All audio files in this dataset are public recordings collected from various sources. The license for each audio file can be found in its parent directory. All derived alignments retain the same licenses as their corresponding audio files. All reference scores are in public domain. All the code is licensed under MIT.
Please cite the following paper if you use the dataset or code provided.
Hao-Wen Dong, Cong Zhou, Taylor Berg-Kirkpatrick, and Julian McAuley, “Deep Performer: Score-to-Audio Music Performance Synthesis,” Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022.