CtrSVDD: A Benchmark Dataset and Baseline Analysis for Controlled Singing Voice Deepfake Detection

TL;DR

This paper introduces CtrSVDD, a comprehensive dataset for singing voice deepfake detection, along with baseline analysis, to improve robustness and generalization of detection models against diverse deepfake methods.

Contribution

The paper provides a large-scale, diverse dataset for singing voice deepfake detection and establishes baseline systems for future research.

Findings

Feature selection significantly impacts detection performance.

Generalization to unseen deepfake methods remains challenging.

The dataset facilitates benchmarking and development of more robust SVDD models.

Abstract

Recent singing voice synthesis and conversion advancements necessitate robust singing voice deepfake detection (SVDD) models. Current SVDD datasets face challenges due to limited controllability, diversity in deepfake methods, and licensing restrictions. Addressing these gaps, we introduce CtrSVDD, a large-scale, diverse collection of bonafide and deepfake singing vocals. These vocals are synthesized using state-of-the-art methods from publicly accessible singing voice datasets. CtrSVDD includes 47.64 hours of bonafide and 260.34 hours of deepfake singing vocals, spanning 14 deepfake methods and involving 164 singer identities. We also present a baseline system with flexible front-end features, evaluated against a structured train/dev/eval split. The experiments show the importance of feature selection and highlight a need for generalization towards deepfake methods that deviate further from training distribution. The CtrSVDD dataset and baselines are publicly accessible.