This is the official implementation for our ISMIR 2022 paper Performance MIDI-to-Score Conversion by Neural Beat Tracking.

The implementation is based on Python 3.8 and PyTorch 1.12.0. Other versions can work, but they are not tested. A list of python dependencies is available in requirements.txt, run:

pip install -r requirements.txt

Use the following code to download the pre-trained models:

wget https://zenodo.org/record/7429669/files/_model_state_dicts.zip
unzip _model_state_dicts.zip

To use the pre-trained model for prediction, please follow the instructions in demo.ipynb. To train a model from scratch, please refer to the instructions in dev/README.md.

L. Liu, Q. Kong, V. Morfi and E. Benetos, "Performance MIDI-to-Score Conversion by Neural Beat Tracking" in Proceedings of the 23rd International Society for Music Information Retrieval Conference, Bengaluru, India, Dec 2022.