A project that trains a LSTM recurrent neural network over a dataset of MIDI files. More information can be found on the writeup about this project or the final report written. Warning: Some parts of this codebase are unfinished.
- Numpy (http://www.numpy.org/)
- Tensorflow (https://github.com/tensorflow/tensorflow)
- Python Midi (https://github.com/vishnubob/python-midi.git)
- Mingus (https://github.com/bspaans/python-mingus)
mkdir data && mkdir models
- Download the dataset Nottingham MIDI dataset and unzip to
data/Nottingham
- Run
python nottingham_util.py
to generate the sequences and chord mapping file todata/nottingham.pickle
- Run
python rnn.py --run_name YOUR_RUN_NAME_HERE
to start training the model. Use the grid object inrnn.py
to edit hyperparameter configurations.