Skip to content

Implementation of the paper "ITO-Master: Inference-Time Optimization for Audio Effects Modeling of Music Mastering Processors"

License

Notifications You must be signed in to change notification settings

SonyResearch/ITO-Master

Repository files navigation

ITO-Master: Inference-Time Optimization for Audio Effects Modeling of Music Mastering Processors

arXiv HuggingFace Audio Samples

This repository contains the official implementation of the paper:

"ITO-Master: Inference-Time Optimization for Audio Effects Modeling of Music Mastering Processors"
Junghyun Koo, Marco A. Martínez-Ramírez, Wei-Hsiang Liao, Giorgio Fabbro, Michele Mancusi, Yuki Mitsufuji
Presented at ISMIR 2025


🔧 Installation

sudo apt-get update && sudo apt-get install -y \
  libsox-fmt-all \
  libsox-dev \
  sox \
  libsndfile1
pip install ito_master

🚀 Inference Examples

Basic usage examples (full examples available in inference_example.sh):

🎛️ Style Transfer Only

python inference.py \
  --input_path examples/input_1.flac \
  --reference_path examples/reference_1.flac \
  --model_type white_box \
  --inference_device cuda \
  --output_dir_path outputs/white_box_st/

🎛️ Style Transfer + ITO (AudioFeatureLoss)

python inference.py \
  --input_path examples/input_1.flac \
  --reference_path examples/reference_1.flac \
  --model_type white_box \
  --inference_device cuda \
  --perform_ito \
  --ito_reference_path examples/reference_1.flac \
  --ito_objective AudioFeatureLoss \
  --num_steps 100 \
  --ito_save_freq 10 \
  --learning_rate 0.01 \
  --output_dir_path outputs/white_box_ito_af/

🎛️ Style Transfer + ITO (CLAPFeatureLoss with text prompt)

python inference.py \
  --input_path examples/input_1.flac \
  --reference_path examples/reference_1.flac \
  --model_type white_box \
  --inference_device cuda \
  --perform_ito \
  --ito_reference_path examples/reference_1.flac \
  --ito_objective CLAPFeatureLoss \
  --clap_target_type Text \
  --clap_text_prompt "heavy metal" \
  --num_steps 100 \
  --ito_save_freq 10 \
  --learning_rate 0.01 \
  --output_dir_path outputs/white_box_ito_claptxt/

📜 Citation

Please cite our work if you find it useful:

@INPROCEEDINGS{koo2025ito, 
  title={ITO-Master: Inference-Time Optimization for Audio Effects Modeling of Music Mastering Processors}, 
  author={Koo, Junghyun and Mart{\'\i}nez-Ram{\'\i}rez, Marco A. and Liao, Wei-Hsiang and Fabbro, Giorgio and Mancusi, Michele and Mitsufuji, Yuki}, 
  booktitle={The 26th International Society for Music Information Retrieval Conference (ISMIR)},  
  year={2025},
}

About

Implementation of the paper "ITO-Master: Inference-Time Optimization for Audio Effects Modeling of Music Mastering Processors"

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published