Skip to content

Latest commit

 

History

History
68 lines (49 loc) · 2.09 KB

README.md

File metadata and controls

68 lines (49 loc) · 2.09 KB
Raw

MuseGAN

PyPI - License PyPI PyPI - Downloads

  • A Pytorch implementation of MuseGAN
  • Note : Only linux BSD support due to SharedArray usage which is a linux only pip package

MuseGAN is a generative model which allows to generate music.

Usage

The musegan package contains all the schema needed to train and generate your own music in MIDI format. There are two ways of using the musegan package;

  1. Using the PYPI package.

    • To use it use the command: pip install musegan to install all the necessary packages

  2. Building from the source

    • First clone the repository to your local directory with git
    • Open your local Terminal/Command Shell and run the following commands
    cd musegan-pytorch # change directory to the cloned repository
#use any of the following some will work dependent on your operating system
#try
python3 setup.py develop
#or
pip install .
#or
pip install -e
    • To test run the following in your python terminal;
    >> import musegan
>> print(musegan__version__)
# which should return the latest musegan version according to the time you read this
0.0.9

    Now you should be able to use zthe musegan packages

Table of content

Training

See this colab notebook for more details of training process.

  • The model components and utils are under musegan/archs folder.
  • The Midi torch.utils.dataset is under musegan/dataset/data_utils.py.
  • The training Functions and criterions can be found in the musegan/trainner folder

License

This project is licensed under MIT.

Links