ICASSP 2026 Automatic Song Aesthetics Evaluation Challenge

dataset·song aesthetics prediction·closed

A completed grand challenge for predicting overall and five fine-grained human aesthetic ratings of AI-generated songs.

Recorded facts

Official site	https://aslp-lab.github.io/Automatic-Song-Aesthetics-Evaluation-Challenge/ ↗
Geography	Global
arxiv	2601.07237
size	tracks: 2399 · expert raters: 16 · duration hours: 140 · challenge tracks: 2
scope	Audio-only prediction of subjective song-aesthetics scores.
genres	nine genres
license	SongEval/challenge-specific terms
steward	ASLP Lab / ICASSP 2026 organizers
creators	ASAE Challenge organizers
languages	English and Chinese songs
modalities	full-song audio; human ratings; challenge splits; leaderboard scores
geographies	Global
access method	Historical challenge website and submission artifacts.
intended uses	song aesthetics prediction; music evaluation research
consent claims	Inherited from SongEval and expert-rater protocol.
related papers	The ICASSP 2026 Automatic Song Aesthetics Evaluation Challenge
annotation method	Expert ratings cover overall musicality plus five fine-grained aesthetic dimensions.
collection method	Uses SongEval as the challenge dataset.
provenance claims	Official challenge and summary paper identify SongEval and task design.
current availability	Challenge concluded; page and paper remain available.
commercial use limits	Dataset rights not established for commercial reuse.
us market scope basis	Publicly available or materially used in US-facing music/audio-AI research.
limitations biases disputes	Aesthetic scores are subjective and culturally contingent.