Python 3 library for converting between MIDI files and Pandas DataFrame objects for use in generative and predictive machine learning applications using symbolic music as training and evaluation material.
Reading and writing to MIDI format is made possible by vishnubob's python-midi project (https://github.com/vishnubob/python-midi). To enable Python 3 support, this dependency must be installed from a specific branch:
pip3 install git+https://github.com/vishnubob/python-midi.git@feature/python3
Once the above requirement has been installed, midi-to-dataframe can be installed as follows:
pip3 install git+https://github.com/TaylorPeer/midi-to-dataframe
Basic usage and configuration is demonstrated in the basics Jupyter notebook: https://github.com/TaylorPeer/midi-to-dataframe/blob/master/examples/basics.ipynb
- Calculation of beat/measure information is incorrect for MIDI files using time signatures other than 4/4.