Song Describer Dataset

dataset·music captions·active·also: SDD

Crowdsourced human descriptions of Creative Commons music for music-language evaluation.

Recorded facts

Official site	https://arxiv.org/abs/2311.10057 ↗
Geography	Global
arxiv	2311.10057
size	captions: 1106 · recordings: 706 · validated captions: 746 · validated recordings: 547
scope	Music captioning, text-to-music and music-language retrieval evaluation.
genres	multiple
license	Per-recording Creative Commons plus dataset release terms
steward	Song Describer research collaboration
creators	Ilaria Manco; Benno Weck; SeungHeon Doh; Minz Won; Yixiao Zhang; Dmitry Bogdanov; Yusong Wu; Ke Chen; Philip Tovstogan; Emmanouil Benetos; Elio Quinton; György Fazekas; Juhan Nam
languages	English captions
modalities	music recordings; English captions; metadata
geographies	Global
access method	Research repository and benchmark task packages.
intended uses	music captioning; text-to-music evaluation; music-language retrieval
related papers	The Song Describer Dataset: a Corpus of Audio Captions for Music-and-Language Evaluation
annotation method	Human-written natural-language descriptions collected by crowdsourcing and validated.
collection method	Creative Commons recordings selected for caption collection.
provenance claims	Paper states all recordings are publicly accessible and released under Creative Commons licenses.
current availability	Dataset and paper available.
commercial use limits	Depends on each source recording's Creative Commons license.
us market scope basis	Publicly available or materially used in US-facing music/audio-AI research.
limitations biases disputes	Small corpus and English-only captions.
redistribution restrictions	Per-recording license and attribution apply.

Current

evaluates with (by)	MuChoMusicSource 1 ↗ VERIFIED high confidence
uses dataset (by)	MIREX 2025 Music Description and CaptioningSource 1 ↗ VERIFIED high confidence