WildSVDD

dataset·in the wild singing deepfake·active

In-the-wild singing voice deepfake detection data extending SingFake for the SVDD 2024 challenge.

Recorded facts

Official site	https://zenodo.org/records/10893604 ↗
Geography	United States
doi	10.5281/zenodo.10893604
size	deposit kb: 657.6 · deposit files: 3
scope	Robust singing-deepfake detection under diverse real-world conditions.
genres	user-generated singing
steward	SVDD Challenge organizers
creators	You Zhang; Yongyi Zang; Jiatong Shi; Ryuichi Yamamoto; Jionghao Han; Yuxun Tang; Tomoki Toda; Zhiyao Duan
versions	v1
languages	multilingual/unspecified
modalities	CSV URL annotations; train split; two test splits
geographies	Global
access method	Open Zenodo annotation files; audio follows referenced sources.
intended uses	singing voice deepfake detection; challenge evaluation
related papers	SVDD 2024: The Inaugural Singing Voice Deepfake Detection Challenge
annotation method	Challenge organizers supply train and test annotations.
collection method	Extends SingFake with updated in-the-wild sources.
provenance claims	Zenodo explicitly identifies WildSVDD as a SingFake extension.
current availability	Annotation files available.
us market scope basis	Publicly available or materially used in US-facing music/audio-AI research.
limitations biases disputes	Deposit primarily contains URL/CSV metadata rather than a durable audio archive.; Challenge report did not provide an official WildSVDD ranking because of variability.
related models tools benchmarks	SingFake; SVDD Challenge 2024