This repository contains a PyTorch implementation of the paper A Repetition-based Triplet Mining Approach for Music Segmentation presented at ISMIR 2023.
The overall format based on the MSAF package.
The network can be trained with:
python trainer.py --feat_id {feature type} --ds_path {path to the dataset}
The dataset format should follow:
dataset/
├── audio # audio files (.mp3, .wav, .aiff)
├── features # feature files (.npy)
└── references # references files (.jams)
To segment tracks and save deep embeddings:
python segment.py --ds_path {path to the dataset} --model_name {trained model name} --bounds {return boundaries and segment labels}
conda env create -f environment.yml
@inproceedings{buisson2023repetition,
title={A Repetition-based Triplet Mining Approach for Music Segmentation},
author={Buisson, Morgan and Mcfee, Brian and Essid, Slim and Crayencour, Helene-Camille},
booktitle={International Society for Music Information Retrieval (ISMIR)},
year={2023}
}