Saved in:
Bibliographic Details
Main Authors: Vohra, Arhan, Akama, Taketo
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.19109
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910001850744832
author Vohra, Arhan
Akama, Taketo
author_facet Vohra, Arhan
Akama, Taketo
contents Perceptual similarity representations enable music retrieval systems to determine which songs sound most similar to listeners. State-of-the-art approaches based on task-specific training via self-supervised metric learning show promising alignment with human judgment, but are difficult to interpret or generalize due to limited dataset availability. We show that pretrained text-audio embeddings (CLAP and MuQ-MuLan) offer comparable perceptual alignment on similarity tasks without any additional fine-tuning. To surpass this baseline, we introduce a novel method to perceptually align pretrained embeddings with source separation and linear optimization on ABX preference data from listening tests. Our model provides interpretable and controllable instrument-wise weights, allowing music producers to retrieve stem-level loops and samples based on mixed reference songs.
format Preprint
id arxiv_https___arxiv_org_abs_2601_19109
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Interpretable and Perceptually-Aligned Music Similarity with Pretrained Embeddings
Vohra, Arhan
Akama, Taketo
Sound
Perceptual similarity representations enable music retrieval systems to determine which songs sound most similar to listeners. State-of-the-art approaches based on task-specific training via self-supervised metric learning show promising alignment with human judgment, but are difficult to interpret or generalize due to limited dataset availability. We show that pretrained text-audio embeddings (CLAP and MuQ-MuLan) offer comparable perceptual alignment on similarity tasks without any additional fine-tuning. To surpass this baseline, we introduce a novel method to perceptually align pretrained embeddings with source separation and linear optimization on ABX preference data from listening tests. Our model provides interpretable and controllable instrument-wise weights, allowing music producers to retrieve stem-level loops and samples based on mixed reference songs.
title Interpretable and Perceptually-Aligned Music Similarity with Pretrained Embeddings
topic Sound
url https://arxiv.org/abs/2601.19109