AFSS: Artifact-Focused Self-Synthesis for Mitigating Bias in Audio Deepfake Detection

TL;DR

This paper introduces AFSS, a novel method that reduces bias in audio deepfake detectors by generating pseudo-fake samples through self-synthesis, improving generalization without relying on pre-collected fake datasets.

Contribution

AFSS is the first approach to use artifact-focused self-synthesis with speaker constraints and dynamic loss reweighting to mitigate bias in deepfake detection.

Findings

Achieves state-of-the-art average EER of 5.45% across 7 datasets.

Significantly reduces EER to 1.23% on WaveFake.

Eliminates the need for pre-collected fake datasets.

Abstract

The rapid advancement of generative models has enabled highly realistic audio deepfakes, yet current detectors suffer from a critical bias problem, leading to poor generalization across unseen datasets. This paper proposes Artifact-Focused Self-Synthesis (AFSS), a method designed to mitigate this bias by generating pseudo-fake samples from real audio via two mechanisms: self-conversion and self-reconstruction. The core insight of AFSS lies in enforcing same-speaker constraints, ensuring that real and pseudo-fake samples share identical speaker identity and semantic content. This forces the detector to focus exclusively on generation artifacts rather than irrelevant confounding factors. Furthermore, we introduce a learnable reweighting loss to dynamically emphasize synthetic samples during training. Extensive experiments across 7 datasets demonstrate that AFSS achieves state-of-the-art performance with an average EER of 5.45\%, including a significant reduction to 1.23\% on WaveFake and 2.70\% on In-the-Wild, all while eliminating the dependency on pre-collected fake datasets. Our code is publicly available at https://github.com/NguyenLeHaiSonGit/AFSS.