Prompt Tuning for Audio Deepfake Detection: Computationally Efficient Test-time Domain Adaptation with Limited Target Dataset

TL;DR

This paper introduces a prompt tuning approach for audio deepfake detection that effectively adapts to new domains with limited data, maintaining high accuracy while being computationally efficient.

Contribution

It presents a novel prompt tuning method for test-time domain adaptation in audio deepfake detection that requires minimal target data and reduces computational costs.

Findings

Improves detection accuracy across different domains.

Requires fewer parameters, suitable for small datasets.

Reduces computational costs compared to traditional methods.

Abstract

We study test-time domain adaptation for audio deepfake detection (ADD), addressing three challenges: (i) source-target domain gaps, (ii) limited target dataset size, and (iii) high computational costs. We propose an ADD method using prompt tuning in a plug-in style. It bridges domain gaps by integrating it seamlessly with state-of-the-art transformer models and/or with other fine-tuning methods, boosting their performance on target data (challenge (i)). In addition, our method can fit small target datasets because it does not require a large number of extra parameters (challenge (ii)). This feature also contributes to computational efficiency, countering the high computational costs typically associated with large-scale pre-trained models in ADD (challenge (iii)). We conclude that prompt tuning for ADD under domain gaps presents a promising avenue for enhancing accuracy with minimal target data and negligible extra computational burden.