TADA: Training-free Attribution and Out-of-Domain Detection of Audio Deepfakes

TL;DR

This paper introduces TADA, a training-free, kNN-based method leveraging a pre-trained SSL model for accurate audio deepfake source attribution and out-of-domain detection, achieving high F1-scores without training.

Contribution

It presents a novel training-free approach for audio deepfake attribution and OOD detection using kNN and SSL models, advancing source tracing capabilities.

Findings

Achieves 0.93 F1-score in source attribution across five datasets.

Demonstrates 0.84 F1-score in out-of-domain detection.

Provides an open-source implementation for reproducibility.

Abstract

Deepfake detection has gained significant attention across audio, text, and image modalities, with high accuracy in distinguishing real from fake. However, identifying the exact source--such as the system or model behind a deepfake--remains a less studied problem. In this paper, we take a significant step forward in audio deepfake model attribution or source tracing by proposing a training-free, green AI approach based entirely on k-Nearest Neighbors (kNN). Leveraging a pre-trained self-supervised learning (SSL) model, we show that grouping samples from the same generator is straightforward--we obtain an 0.93 F1-score across five deepfake datasets. The method also demonstrates strong out-of-domain (OOD) detection, effectively identifying samples from unseen models at an F1-score of 0.84. We further analyse these results in a multi-dimensional approach and provide additional insights. All code and data protocols used in this work are available in our open repository: https://github.com/adrianastan/tada/.