Search papers, labs, and topics across Lattice.
This paper introduces MADB, a large-scale music aesthetics dataset consisting of 9,999 tracks annotated by 30 trained annotators across 10 perceptual dimensions and an overall score. The dataset addresses the critical gap in structured aesthetic annotations, enabling a comprehensive evaluation of models in music aesthetic assessment. Findings indicate significant discrepancies between model predictions and human judgments, highlighting the limitations of existing approaches in capturing nuanced human perceptions of music.
Current models struggle to align with human music aesthetic judgments, revealing a substantial gap in understanding that MADB aims to bridge.
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
