Description

This is the official repository for the paper "MusicMagus: Zero-Shot Text-to-Music Editing via Diffusion Models".

If there is any problem related to the code running, please open an issue and I will help you as mush as I can.

Demo page

https://bit.ly/musicmagus-demo

Installation

Pip

# clone project
git clone https://github.com/ldzhangyx/MusicMagus/
cd MusicMagus

# [OPTIONAL] create conda environment
conda create -n myenv python=3.11.7
conda activate myenv

# install pytorch according to instructions
# https://pytorch.org/get-started/

# install requirements
pip install -r requirements.txt

Conda

# clone project
git clone https://github.com/ldzhangyx/MusicMagus/
cd MusicMagus

# create conda environment and install dependencies
conda env create -f environment.yaml -n myenv

# activate conda environment
conda activate myenv

Configuring LP-MusicCaps

Please download transfer.pth from the website below and place it to lpmc/music_captioning/exp/transfer/lp_music_caps/ folder.

https://huggingface.co/seungheondoh/lp-music-caps/blob/main/transfer.pth

How to run

1. Set the openai.key value in audioldm2/embedding_calculator.py.

2. Directly run inference.ipynb.

Citation

@misc{zhang2024musicmagus,
      title={MusicMagus: Zero-Shot Text-to-Music Editing via Diffusion Models}, 
      author={Yixiao Zhang and Yukara Ikemiya and Gus Xia and Naoki Murata and Marco A. Martínez-Ramírez and Wei-Hsiang Liao and Yuki Mitsufuji and Simon Dixon},
      year={2024},
      eprint={2402.06178},
      archivePrefix={arXiv},
      primaryClass={cs.SD}
}