Authors

Venue

Date

Share

Discogs-VINet-MIREX

Xavier Serra

Yuki Mitsufuji

R.O. Araz

J. Serrà

D. Bogdanov

MIREX 2024

2025

Abstract

This technical report presents our submission to the cover song identification task for the 2024 edition of the Music Information Retrieval Evaluation eXchange (MIREX). For this submission, we enhanced our Discogs-VINet model by changing the definition of an epoch, incorporating automatic mixed precision (AMP) during both training and inference, and sampling four versions per clique during triplet mining (which became possible with AMP). Due to this enhanced model’s performance on the Discogs-VI test set, we trained a new model from scratch using the entire Discogs-VI dataset, rather than just the training partition used in Discogs-VINet (a 45% increase in the number of versions). This enhanced and retrained model is named Discogs-VINet-MIREX.

Related Publications

Training Consistency Models with Variational Noise Coupling

ICML, 2025
Gianluigi Silvestri, Luca Ambrogioni, Chieh-Hsin Lai, Yuhta Takida, Yuki Mitsufuji

Consistency Training (CT) has recently emerged as a promising alternative to diffusion models, achieving competitive performance in image generation tasks. However, non-distillation consistency training often suffers from high variance and instability, and analyzing and impr…

Supervised Contrastive Learning from Weakly-labeled Audio Segments for Musical Version Matching

ICML, 2025
Joan Serrà, R. Oguz Araz, Dmitry Bogdanov, Yuki Mitsufuji

Detecting musical versions (different renditions of the same piece) is a challenging task with important applications. Because of the ground truth nature, existing approaches match musical versions at the track level (e.g., whole song). However, most applications require to …

Distillation of Discrete Diffusion through Dimensional Correlations

ICML, 2025
Satoshi Hayakawa, Yuhta Takida, Masaaki Imaizumi*, Hiromi Wakaki*, Yuki Mitsufuji

Diffusion models have demonstrated exceptional performances in various fields of generative modeling, but suffer from slow sampling speed due to their iterative nature. While this issue is being addressed in continuous domains, discrete diffusion models face unique challenge…

JOIN US

Shape the Future of AI with Sony AI

We want to hear from those of you who have a strong desire
to shape the future of AI.