From Sharpness to Better Generalization for Speech Deepfake Detection

TL;DR

This paper explores the role of sharpness as a theoretical measure of generalization in speech deepfake detection, demonstrating that reducing sharpness improves robustness across unseen conditions.

Contribution

It introduces sharpness as a theoretical proxy for generalization in SDD and applies sharpness-aware minimization to enhance model robustness.

Findings

Sharpness increases under domain shifts, indicating higher sensitivity.

Reducing sharpness improves performance on unseen test sets.

Sharpness correlates significantly with generalization in most settings.

Abstract

Generalization remains a critical challenge in speech deepfake detection (SDD). While various approaches aim to improve robustness, generalization is typically assessed through performance metrics like equal error rate without a theoretical framework to explain model performance. This work investigates sharpness as a theoretical proxy for generalization in SDD. We analyze how sharpness responds to domain shifts and find it increases in unseen conditions, indicating higher model sensitivity. Based on this, we apply Sharpness-Aware Minimization (SAM) to reduce sharpness explicitly, leading to better and more stable performance across diverse unseen test sets. Furthermore, correlation analysis confirms a statistically significant relationship between sharpness and generalization in most test settings. These findings suggest that sharpness can serve as a theoretical indicator for generalization in SDD and that sharpness-aware training offers a promising strategy for improving robustness.