Melody-and-Sound-Generation

Final project of the course "Implementing artificial neural networks with Tensorflow"

Sound Generation

To train the variational autoencoder, you have to retrieve the dataset from https://magenta.tensorflow.org/datasets/nsynth#files and copy it to the folder 'Sound Generation\dataset\fsdd\recordings'. Then run preprocess.py Then run the respective cells in music_vae.ipynb

Melody Generation

You can find related datasets from http://www.piano-midi.de