CtrSVDD

dataset·controlled singing deepfake·active

Controlled singing-voice deepfake corpus generated from public singing data with synthesis and voice-conversion systems.

Recorded facts

Official site	https://github.com/SVDDChallenge/CtrSVDD ↗
Geography	Global
github repository	SVDDChallenge/CtrSVDD
size	exact total: see official evaluation plan
scope	Controlled singing voice deepfake detection and anti-spoofing.
genres	singing voice
steward	SVDD Challenge organizers
creators	You Zhang; Yongyi Zang; Jiatong Shi; Ryuichi Yamamoto; Tomoki Toda; Zhiyao Duan
languages	multilingual/unspecified
modalities	audio; real/fake labels; baseline code; generation-system metadata
geographies	Global
access method	Official GitHub repository and challenge resources.
intended uses	singing voice deepfake detection; anti-spoofing evaluation
related papers	CtrSVDD: A Benchmark Dataset and Baseline Analysis; SVDD 2024
annotation method	Known generation pipeline supplies controlled authenticity labels.
collection method	Public singing-vocal data transformed with singing synthesis and conversion systems.
provenance claims	Challenge paper describes use of publicly available singing-vocal data.
current availability	Dataset and baseline described as available on challenge site.
us market scope basis	Publicly available or materially used in US-facing music/audio-AI research.
limitations biases disputes	Controlled generators may not represent novel or in-the-wild systems.; Voice and repertoire coverage depends on upstream public singing data.
related models tools benchmarks	SVDD Challenge 2024