Stellar spectra-driven unsupervised music composition system.
CONTENTS
- Jupyter notebook 1-DeepVAE_4dim.ipynb
- Jupyter notebook 2-Latent2Chords_4dim.ipynb
- Jupyter notebook 3-MAESTRO2numpy.ipynb
- Jupyter notebook 4-AI-rmonizer_Demo.ipynb
- LICENSE
- README
- requirements.txt
AI-RMONIZER INSTALLATION
1 - Download all the content of this repository into the same folder
2 - Install all the dependencies included in the requirements.txt file
Generating stellar score:
1 - Download the stellar spectra from: http://svocats.cab.inta-csic.es/miles/index.php?action=credits
2 - Open the Jupyter notebook 1-DeepVAE_4dim.ipynb, and introduce your paths to the downloaded spectra
3 - Run all cells in the Jupyter notebook
Preprocessing:
1 - Run the Jupyter notebook 2-Latent2Chords_4dim.ipynb
2 - Download the MAESTRO dataset from: https://magenta.tensorflow.org/datasets/maestro
3 - Open the Jupyter notebook 3-MAESTRO2numpy.ipynb, and introduce your paths to the downloaded MIDI files
4 - Run all cells in the Jupyter notebook
Generating the final spectra-driven score:
1 - Download the pre-trained weights from Zenodo : https://zenodo.org/record/8342535
2 - Download the MIDI file: http://www.jsbach.net/midi/midi_solo_cello.html
3 - Open the Jupyter notebook 4-AI-rmonizer_Demo.ipynb, and introduce your paths to the downloaded files
4- Run all cells in the Jupyter notebook
Enjoy the piece!
Developer system info: Python 3.8.5 - Cabbage 2.5.0 - i7 macOS 10.15.7 - 32 GB - 1536 MB