Automatic Music Sample Identification with Multi-Track Contrastive Learning

Alain Riou, Joan Serrà, Yuki Mitsufuji
10/13/2025
cs.SDcs.AIcs.LG

No official code found. Best implementation: sony/sampleid (27 stars, MIT, PyTorch, last commit 2026-01-28).

Abstract

Sampling, the technique of reusing pieces of existing audio tracks to create new music content, is a very common practice in modern music production. In this paper, we tackle the challenging task of automatic sample identification, that is, detecting such sampled content and retrieving the material from which it originates. To do so, we adopt a self-supervised learning approach that leverages a multi-track dataset to create positive pairs of artificial mixes, and design a novel contrastive learning objective. We show that such method significantly outperforms previous state-of-the-art baselines, that is robust to various genres, and that scales well when increasing the number of noise songs in the reference database. In addition, we extensively analyze the contribution of the different components of our training pipeline and highlight, in particular, the need for high-quality separated stems for this task.

View on arXivView PDF

Code Implementations(1)

sony/sampleidCommunity65%
PyTorchMIT

Code for the paper “Automatic Music Sample Identification with Multi-Track Contrastive Learning”.

273Shell, PythonOct 15, 20258 months agoMIT
audiocontrastive-learningdeep-learningmachine-learningmusic+6 more

Similar Papers

Refining music sample identification with a self-supervised graph neural network

83% match1 repo192025

An Experimental Comparison Of Multi-view Self-supervised Methods For Music Tagging

83% match1 repo72024

Self-Supervised Contrastive Learning for Robust Audio-Sheet Music Retrieval Systems

82% match2023

Contrastive timbre representations for musical instrument and synthesizer retrieval

80% match2025

Contrastive Learning from Synthetic Audio Doppelgängers

80% match1 repo42024

Self-refining of Pseudo Labels for Music Source Separation with Noisy Labeled Data

80% match2023

Cite this paper

@article{riou2025automatic,
  title  = {Automatic Music Sample Identification with Multi-Track Contrastive Learning},
  author = {Alain Riou and Joan Serrà and Yuki Mitsufuji},
  year   = {2025},
  eprint = {2510.11507},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2510.11507v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2510.11507.svg)](https://codeofpaper.com/paper/2510.11507)

Discussion