Melody Maker

Melody Maker is a machine-learning driven music generation toolkit developed by Waltworks. It leverages RNN models to learn patterns from MIDI datasets and generate new melodies and chord progressions.

---

Table of Contents

1. Features
2. Prerequisites
3. Installation
4. Configuration
5. Usage
   • Training
   • Generation
6. Project Structure
7. Contributing
8. License

---

Features

• Data preprocessing: Convert MIDI files to note-state matrices.
• Model training: Train melody and chord RNN models on your MIDI dataset.
• Music generation: Produce new MIDI files from trained models.
• Configurable: Easily adjust sequence length, model hyperparameters, and data directories.

---

Prerequisites

• Python 3.8+
• make (GNU Make)
• Virtual environment tool (e.g., venv, conda)

Python Libraries

numpy
mido
tensorflow>=2.6
scikit-learn
tqdm

---

Installation

1. Clone the repository

   git clone https://github.com/your-org/melody-maker.git
   cd melody-maker

2. Create and activate a virtual environment

   python3 -m venv .venv
   source .venv/bin/activate

3. Install dependencies

   pip install -r requirements.txt

---

Configuration

All configurable parameters live in src/config.py. Key settings include:

Variable	Description	Default
MIDI_DIR	Path to your raw MIDI training files	Rock_Music_Midi
LOWER_BOUND	Lowest MIDI pitch to include	24
UPPER_BOUND	Highest MIDI pitch to include	102
SEQ_LEN	Sequence length (timesteps)	60
BATCH_SIZE	Training batch size	64
EPOCHS	Number of training epochs	20
MELODY_MODEL_PATH	Output path for melody model checkpoint	melody_model.h5
CHORD_MODEL_PATH	Output path for chord model checkpoint	chord_model.h5

---

Usage

Training

Note: You only need to run training once per dataset.

make train

This will:

1. Preprocess all MIDI files in MIDI_DIR.
2. Train the melody RNN model (saved to MELODY_MODEL_PATH).
3. Train the chord RNN model (saved to CHORD_MODEL_PATH).

Generation

You can regenerate music at any time once models are trained:

make generate

This will:

1. Load the trained models.
2. Generate a new melody and chord sequence.
3. Export a MIDI file named output_generated.mid.

---

Project Structure

.
├── Makefile              # Defines `train` and `generate` targets
├── requirements.txt      # Python dependencies
├── README.md             # This document
└── src
    ├── config.py         # Configuration parameters
    ├── data_utils.py     # Encoders and scalers
    ├── midi_tools.py     # MIDI ↔ state-matrix conversion
    ├── rnn.py            # Model definitions & generation logic
    └── generate.py       # Entry point for training & generation

---

Contributing

We welcome contributions! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/YourFeature)
3. Commit your changes (git commit -m "Add your feature")
4. Push to the branch (git push origin feature/YourFeature)
5. Open a Pull Request

Please adhere to the existing code style and include tests for new functionality.

---

License

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