Few-Shot Speech Deepfake Detection Adaptation with Gaussian Processes

TL;DR

This paper presents ADD-GP, a Gaussian Process-based few-shot adaptive framework for audio deepfake detection that effectively adapts to unseen TTS models and supports personalized detection with minimal data.

Contribution

The paper introduces ADD-GP, a novel few-shot adaptive detection framework combining deep embeddings with Gaussian Processes for improved robustness and adaptability.

Findings

Achieves strong performance in detecting unseen deepfake models.

Demonstrates effective one-shot personalization of detection.

Provides a new benchmark dataset for evaluation.

Abstract

Recent advancements in Text-to-Speech (TTS) models, particularly in voice cloning, have intensified the demand for adaptable and efficient deepfake detection methods. As TTS systems continue to evolve, detection models must be able to efficiently adapt to previously unseen generation models with minimal data. This paper introduces ADD-GP, a few-shot adaptive framework based on a Gaussian Process (GP) classifier for Audio Deepfake Detection (ADD). We show how the combination of a powerful deep embedding model with the Gaussian processes flexibility can achieve strong performance and adaptability. Additionally, we show this approach can also be used for personalized detection, with greater robustness to new TTS models and one-shot adaptability. To support our evaluation, a benchmark dataset is constructed for this task using new state-of-the-art voice cloning models.