/DeepJ

A DeepJ variant trained to produce songs conforming to an emotion.

Primary LanguageJupyter NotebookMIT LicenseMIT

DeepJ-Mood: A variant of the DeepJ model for mood-specific music generation

Sample Outputs

Listen to our best outputs at

Code

A COE 197-Z / EE 298-Z Final Project.

References

Usage

To train a new model, run the following command:

python train.py

To generate music, run the following command:

python generate.py

Use the help command to see CLI arguments:

python generate.py --help

Piano2Vec : A word2vec model trained on the harmonic reduction of 140 piano midi files from MAESTRO Dataset

To use the word2vec model, run with word2vec folder in the same directory

python midi-similarity.py