thekc33 / melodia

Python-based tool for generating musical compositions using algorithmic techniques.

Melodia: Algorithmic Music Composition in Python

Melodia is a Python-based tool for generating musical compositions using algorithmic techniques. It uses machine learning and deep learning techniques to create unique and creative musical pieces. Explore the creative possibilities of code and music!

CURRENTLY WORKING ON:

• Multi-track generation and multi hand piano support

Features

• Generate unique musical compositions using your own MIDI files.
• Uses LSTM (Long Short-Term Memory) networks, a type of recurrent neural network, to learn musical patterns and generate new ones.
• Save and load trained models for later use.
• Export generated compositions as MIDI files.

Installation

1. Clone this repository: git clone https://github.com/TheKC33/melodia.git
2. Navigate to the cloned repository: cd melodia
3. Install dependencies: pip install -r requirements.txt

Usage

1. Prepare your example MIDI files and place them in the examples directory.
2. Run the training script: python train.py
3. After training, go into the play.py and change the model name to the model you just trained.
4. Run the play script to generate a new composition: python play.py
5. The generated composition will be saved as a MIDI file.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgements

The data used to train this model is built upon the MAESTRO dataset, which was introduced in the following paper:

Curtis Hawthorne, Andriy Stasyuk, Adam Roberts, Ian Simon, Cheng-Zhi Anna Huang, Sander Dieleman, Erich Elsen, Jesse Engel, and Douglas Eck. "Enabling Factorized Piano Music Modeling and Generation with the MAESTRO Dataset." In International Conference on Learning Representations, 2019.

The MAESTRO dataset can be found at: https://magenta.tensorflow.org/datasets/maestro