Rag-MusicPrompt

Rag-MusicPrompt is a Python-based project that demonstrates how to generate music from natural language prompts using Hugging Face's MusicGen model.

It includes utilities for audio export, retro effects (bitcrusher + downsampling), and reproducibility with seeds.

This repository can be used both as a tool for experimentation and as a foundation for integrating AI-generated music into indie games, multimedia projects, or research.

---

🚀 Features

• Text-to-audio generation using prompts (style, instruments, mood, etc.).
• Adjustable guidance / sampling for creative fidelity vs consistency.
• Retro mode: bit-crushing + downsampling + optional aliasing, to get 8-bit / lo-fi game vibes.
• Output in WAV and MP3 formats.
• Command-line interface; configurable via arguments.
• Useful for indie game music assets: menus, level intros, ambient loops, etc.,
• GPU support (CUDA).

---

📂 Repository Structure

Rag-MusicPrompt/
├── main.py              # Main script with CLI
├── README.md            # Project documentation
├── requirements.txt     # Python dependencies (optional)
├── /output              # Default output folder (created automatically)

---

📦 Installation

It is recommended to use a virtual environment to isolate dependencies.

1. Clone the repository

git clone https://github.com/RegoDefies/Rag-MusicPrompt.git
cd Rag-MusicPrompt

2. Create and activate a virtual environment

Linux / macOS:

python3 -m venv venv
source venv/bin/activate

Windows:

python -m venv venv
venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

If requirements.txt is not available, install directly:

pip install torch transformers pydub soundfile numpy

---

🚀 Usage

Run the tool from the command line with a text prompt:

python main.py --prompt "lofi hip hop beat with piano and rain sounds" --seconds 15 --out output_folder

Arguments

• --prompt (str, required): Music description prompt.
• --seconds (int, default=10): Duration in seconds.
• --seed (int): Random seed for reproducibility.
• --guidance (float, default=7.5): Prompt fidelity.
• --sample / --no-sample: Enable or disable sampling.
• --device (str): Device (e.g., "cuda:0" for GPU).
• --out (str, default="output"): Output folder.
• --basename (str, default="music"): Base filename.
• --bitrate (str, default="192k"): MP3 bitrate.

Retro Mode Options

• --retro: Enable retro mode.
• --bitdepth (int, default=8): Bitcrusher depth.
• --sr_target (int, default=11025): Target sample rate.
• --no-aa: Disable anti-alias filter.

---

🎼 Examples

1. Epic orchestral soundtrack

python main.py --prompt "Epic orchestral soundtrack with choir" --seconds 20 --out music_out --basename epic_track

2. 8-bit retro game vibe

python main.py --prompt "8-bit retro game music, fast tempo" --seconds 12 --retro --bitdepth 6 --sr_target 8000 --out retro_out --basename chip_track

3. Chill lofi loop

python main.py --prompt "Chill lofi hip hop with vinyl crackle" --seconds 30 --out lofi_out --basename lofi_track

---

📜 License

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