SDLGOct 6, 2025

A Study on the Data Distribution Gap in Music Emotion Recognition

arXiv:2510.04688v11 citationsh-index: 4
Originality Incremental advance
AI Analysis

This addresses dataset bias issues in music emotion recognition for applications requiring robust models across diverse musical styles.

The paper tackled the problem of out-of-distribution generalization in Music Emotion Recognition by investigating five datasets spanning various musical genres, and proposed a framework combining Jukebox embeddings with chroma features that substantially improved cross-dataset generalization capabilities.

Music Emotion Recognition (MER) is a task deeply connected to human perception, relying heavily on subjective annotations collected from contributors. Prior studies tend to focus on specific musical styles rather than incorporating a diverse range of genres, such as rock and classical, within a single framework. In this paper, we address the task of recognizing emotion from audio content by investigating five datasets with dimensional emotion annotations -- EmoMusic, DEAM, PMEmo, WTC, and WCMED -- which span various musical styles. We demonstrate the problem of out-of-distribution generalization in a systematic experiment. By closely looking at multiple data and feature sets, we provide insight into genre-emotion relationships in existing data and examine potential genre dominance and dataset biases in certain feature representations. Based on these experiments, we arrive at a simple yet effective framework that combines embeddings extracted from the Jukebox model with chroma features and demonstrate how, alongside a combination of several diverse training sets, this permits us to train models with substantially improved cross-dataset generalization capabilities.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes