FakeMusicCaps

dataset·synthetic music detection·active

MusicCaps prompts regenerated with multiple text-to-music systems for synthetic-music detection and generator attribution.

Verified facts

Official site	https://zenodo.org/records/15063698 ↗
Geography	Italy
doi	10.5281/zenodo.15063698
arxiv	2409.10684
github repository	polimi-ispl/FakeMusicCaps
size	generated samples: 27605
scope	Detection and attribution of text-to-music synthetic audio.
genres	MusicCaps-derived mix
steward	Politecnico di Milano ISPL
creators	Luca Comanducci; Paolo Bestagini; Stefano Tubaro
languages	English prompts
modalities	generated audio; MusicCaps captions/prompts; generator labels
geographies	Global
access method	Zenodo dataset and official GitHub code.
intended uses	synthetic music detection; generator attribution; open-set forensics
related papers	FakeMusicCaps: a Dataset for Detection and Attribution of Synthetic Music
annotation method	Generator identity supplies synthetic and attribution labels.
collection method	Multiple text-to-music models regenerate MusicCaps music-caption pairs.
provenance claims	Derived from MusicCaps captions and generated by named text-to-music systems.
current availability	Dataset and code available.
us market scope basis	Publicly available or materially used in US-facing music/audio-AI research.
limitations biases disputes	Closed-set detection may learn generator-specific artifacts.; MusicCaps prompts and source distribution constrain coverage.
related models tools benchmarks	MusicCaps; M6