Saved in:
Bibliographic Details
Main Authors: Roy, Abhinaba, Liang, Junyi, Herremans, Dorien
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.27346
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • Current music similarity models typically compute a single, monolithic score, entangling distinct musical dimensions like melody, rhythm, and timbre. This limits user control and interpretability, making it impossible to execute nuanced queries. We introduce MERIT, a framework for learning disentangled, factor-specific music representations tailored to these three core dimensions. To overcome the lack of isolated musical variations in real-world audio, we use a novel training strategy that uses conditional audio generation and source-separated stems to strongly encourage single-factor variation in training data. Our evaluations demonstrate strong factor-wise disentanglement. Each head responds strongly to its intended perceptual dimension while remaining near chance on the others, a representational property that holds across both the synthetic training domain and independent real-world audio.