The Urgent Need for Transdisciplinary Approaches to Audio Deepfake Detection and Discernment
Current AI-based methods for detecting audio deepfakes are insufficient due to their reliance on limited training data and lack of understanding of the complexities of human language; a transdisciplinary approach incorporating linguistic expertise and human discernment is crucial for more robust and comprehensive detection.
Prompt Tuning for Audio Deepfake Detection: Enhancing Accuracy with Limited Target Data and Minimal Computational Overhead
Prompt tuning, a parameter-efficient fine-tuning method, effectively adapts pre-trained audio deepfake detection models to new domains with limited data, addressing challenges of domain gaps, data scarcity, and computational costs.
Explainable AI for Detecting Real-World Audio Deepfakes: A New Benchmark for Generalizability
Current AI-based audio deepfake detection methods, while effective in controlled settings, often fail to generalize to real-world scenarios and lack the transparency needed to foster trust with users. This paper introduces a novel benchmark for evaluating the generalizability of these methods and explores explainability techniques to bridge the gap between performance and user understanding.
SONAR: A Benchmark for Detecting Audio Deepfakes Generated by Cutting-Edge AI
Existing audio deepfake detection models struggle to generalize across diverse datasets and against advanced text-to-speech (TTS) models, highlighting the need for more robust detection methods and comprehensive benchmarks like SONAR.
Diffusion and Flow-Matching Based Audio Deepfake Dataset: Assessing the Robustness of Anti-Spoofing Models Against Advanced Text-to-Speech Synthesis
The Diffusion and Flow-Matching Based Audio Deepfake (DFADD) dataset provides a comprehensive collection of spoofed audio generated by state-of-the-art diffusion and flow-matching text-to-speech (TTS) models, enabling the development of more robust anti-spoofing detection models.
Robust Transformer-based Audio Deepfake Detection with Continuous Learning Capabilities
This paper proposes a novel framework for audio deepfake detection that achieves high accuracy on available fake data and effectively performs continuous learning on new fake data using few-shot learning.
Enhancing Generalization in Audio Deepfake Detection: A Neural Collapse-based Sampling and Training Approach
A neural collapse-based sampling approach to create a new training database from diverse datasets, enabling computationally efficient and generalized audio deepfake detection models.
Advancing Audio Deepfake Detection: A Novel Cross-Domain Dataset and Comprehensive Analysis
This paper presents a novel cross-domain audio deepfake detection (CD-ADD) dataset comprising over 300 hours of speech data generated by five advanced zero-shot text-to-speech (TTS) models. The dataset is designed to simulate real-world scenarios and evaluate the generalization capabilities of deepfake detection models.
