Singing Voice Deepfake Detection (SVDD)

The SVDD task aims to detect AI-generated singing voices, which is an emerging issue within the music industry that requires specialized solutions.

Challenges

The SVDD task aims to detect AI-generated singing voices, which is an emerging issue within the music industry that requires specialized solutions.

WildSVDD

@MIREX Challenge 2024

Detect AI-generated singing voices in the wild, addressing real-world scenarios and diverse audio conditions.

Deadline: Oct 10, 2024 (TBD) Open for Submissions

SVDD Challenge

@IEEE SLT Workshop 2024

Develop advanced algorithms for singing voice deepfake detection, pushing the boundaries of audio forensics.

Deadline: Closed Submissions Closed

Datasets and Publications

Explore our datasets and publications related to Singing Voice Deepfake Detection.

WildSVDD

Recently Updated

Newly updated from SingFake dataset for singing voice deepfake detection in the wild.

Status: Updated New Dataset

CtrSVDD

@INTERSPEECH 2024

Includes baseline system and dataset for singing voice deepfake detection.

Status: Available Dataset & Baseline

SingFake

@ICASSP 2024

Provides baseline system and dataset for singing voice deepfake detection.

Status: Available Dataset & Baseline

Organizers

You Zhang

You Zhang

University of Rochester

Yongyi Zang

Yongyi Zang

University of Rochester

Jiatong Shi

Jiatong Shi

Carnegie Mellon University

Ryuichi Yamamoto

Ryuichi Yamamoto

Nagoya University

Tomoki Toda

Tomoki Toda

Nagoya University

Zhiyao Duan

Zhiyao Duan

University of Rochester