LSTM Music Generator

About

Machine Learning project to generate music using LSTM. Currently, it is being trained on Midi files of Super Mario Bros 2

Running Program

You may use Main.py --help to see all the options

Optional Arguments:
  -h, --help            Show this help message and exit
  -M MODEL, --Model MODEL
                        Model A or B (Default: Model 'A')
  -W WEIGHTS, --Weights WEIGHTS
                        Load weights to continue training (Default: None)
  -D DIRECTORY, --Directory DIRECTORY
                        Directory of Midi Files (Default: '.')
  -E EPOCHS, --Epochs EPOCHS
                        Number of Epochs (Default: 100)
  -O OUTPUTS, --Outputs OUTPUTS
                        Number of Generated Output(s) (Default: 1)
  -BS BATCH_SIZE, --Batch_Size BATCH_SIZE
                        Batch Size (Default: 128)
  -SL SEQUENCE_LENGTH, --Sequence_Length SEQUENCE_LENGTH
                        Sequence Length (Default: 100)

• There are two models, A and B. Checkout code to see the difference.
• Weights are optional to load. If you want to continue training from certain weights, you may use this feature.
• Set a directory where the MIDI file are contained.
• Set the number of Epochs, the default is 10.
• The Output is how many sound that will be generated after training. We recommend >= 3
• Set Batch Size.
• Set Sequence Length.

All of these options have a set default.

Example Runs

Main.py --Model B --Directory Undertale --Epochs 10 --Outputs 5 --Batch_Size 128 --Sequence_Length 80

Recommendations

• We strongly recommend for the user to run this code and train on Anaconda, due to a drastic performance increase while using Anaconda.

Contributors

• Yousif Kako
• Faraz Heravi
• Dr. Andrew Rosen