Make sure that requirements in requirements.txt
are installed, or run
pip install -r requirements.txt
Then, make sure you have FluidSynth and a .sf2
soundfont installed.
- Create a file named
.env
in the project's root directory, following the template shown in the.env.example
file. - Execute
processing/preprocess_batch.py
using Python. You must have the dataset and sufficient disk space of [] MB to store the preprocessed data. If you wish to only preprocess a subset, specify the--year
argument.
- Dataloaders (for pytorch) for all components of the dataset is located in model/database.py. Use this to load your data.
- Then, you can train the models we have built using the
fit
method, and evaluate them using theval_split
method. To use your own models, you can still use the dataloaders.
- You can use the code in this jupyter notebook to make predictions. However, ensure you have trained some sort of model to make the predictions.
If there are any new datasets added, please update the README with the file structures.
- MAESTRO should look like
.
├── 2004
├── 2006
├── 2008
├── 2009
├── 2011
├── 2013
├── 2014
├── 2015
├── 2017
├── 2018
├── LICENSE
├── maestro-v3.0.0.csv
├── maestro-v3.0.0.json
└── README
- MusicNet should look like
.
├── test_data
├── test_labels
├── train_data
└── train_labels