MADB: A Large-Scale Music Aesthetics Dataset with Professional and Multi-Dimensional Annotations

2026-07-07

AI Summary

This study addresses the scarcity of large-scale, structured, and multidimensional human perceptual annotations in music aesthetic evaluation, which hinders effective modeling of human aesthetic judgments. To bridge this gap, the authors construct a high-quality dataset comprising 9,999 musical pieces, annotated by 30 professional raters across ten perceptual dimensions and overall aesthetic quality, supplemented with textual comments to enable multimodal analysis. The work also introduces a unified model evaluation framework. This dataset represents the first integration of large-scale, expert-level, multidimensional ratings with free-text commentary, establishing a new benchmark for human-aligned music understanding. Empirical evaluations reveal a significant discrepancy between current computational models and human aesthetic judgments, highlighting critical limitations in existing approaches.

Abstract

Music aesthetic assessment is a challenging yet underexplored problem, requiring models to capture fine-grained, multi-dimensional human perceptual judgments. Progress in this area has been limited by the lack of large-scale datasets with structured aesthetic annotations. We introduce MADB, a large-scale dataset and benchmark comprising 9,999 tracks annotated by 30 trained annotators. Each track is rated by around 10 annotators across 10 perceptual dimensions and one overall score, with additional textual comments for multimodal analysis. We establish a unified evaluation framework over multiple pretrained models. Results reveal substantial gaps between model predictions and human judgments, exposing key limitations of current approaches. MADB provides a new benchmark for human-aligned music understanding. Project page: https://github.com/knownree/madb

Problem

• music aesthetic assessment
• large-scale dataset
• multi-dimensional annotations
• human perceptual judgments
• structured aesthetic annotations

Innovation

• music aesthetics
• large-scale dataset
• multi-dimensional annotation
• human-aligned evaluation
• multimodal analysis