lofi-gen 0.1.0

A Python package for generating Lofi music using AI models

lofi-gen

A Python package for generating Lofi music using AI models like MusicGen.

Features

• Generate lofi music from text prompts using Meta's MusicGen model
• Create seamless long-form music loops (extend short clips to hours-long sessions)
• Easy-to-use API for music generation
• Spectral crossfading for smooth transitions

Installation

From PyPI

pip install lofi-gen

From Source

git clone https://github.com/gnikesh/lofi-gen.git
cd lofi-gen
pip install -e .

Quick Start

Basic Music Generation

from lofi_gen.music.models import MusicGenModel

# Initialize the model
model = MusicGenModel(model_size="large")  # Options: "small", "medium", "large"

# Generate music from a text prompt
audio, sample_rate = model.generate_music(
    prompt="lofi hip hop beat with soft piano, chill vibes",
    duration_seconds=30
)

# Save the generated audio
model.save_audio(audio, sample_rate, "output.wav")

Generate Long-Form Music

from lofi_gen.music.models import MusicGenModel
from lofi_gen.music.pipelines import LongMusicGenerator

# Generate initial clip
model = MusicGenModel()
audio, sample_rate = model.generate_music(
    prompt="relaxing nepali lofi with bansuri flute",
    duration_seconds=60
)
model.save_audio(audio, sample_rate, "base_clip.wav")

# Extend it to a long seamless loop
extender = LongMusicGenerator(crossfade_sec=3.0, variation=True)
extender.generate(
    input_file="base_clip.wav",
    output_file="extended_lofi.wav",
    target_duration_mins=60  # Create a 60-minute loop
)

Requirements

• Python >= 3.8
• PyTorch >= 2.0.0
• transformers >= 4.30.0
• librosa >= 0.10.0
• scipy, numpy, soundfile

See requirements.txt for full dependencies.

Model Sizes

MusicGen comes in three sizes:

• small: Fastest, lower quality (~300MB)
• medium: Balanced speed and quality (~1.5GB)
• large: Best quality, slower (~3.3GB)

Choose based on your hardware and quality requirements.

Advanced Usage

Custom Generation Parameters

audio, sample_rate = model.generate_music(
    prompt="lofi hip hop beat",
    duration_seconds=30,
    guidance_scale=3.0,      # How closely to follow the prompt (1.0-15.0)
    temperature=1.0,         # Randomness (0.1-2.0)
    top_k=250,              # Sampling diversity
    top_p=0.0               # Nucleus sampling (0 to disable)
)

Seamless Loop Configuration

extender = LongMusicGenerator(
    crossfade_sec=3.0,    # Crossfade duration
    variation=True        # Add subtle variations to prevent monotony
)

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

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

License

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

Acknowledgments

• Meta AI for the MusicGen model
• Hugging Face for the Transformers library
• The open-source community for various audio processing tools

Support

For issues, questions, or suggestions:

• Open an issue on GitHub

Roadmap

• Add more AI models (AudioLDM, Stable Audio, etc.)
• Video generation capabilities
• CLI interface for easy command-line usage
• Pre-trained models fine-tuned on Nepali music
• Web interface for music generation