A comprehensive survey of audio forgery detection: challenges and novel trends

Abstract The growing accessibility of advanced audio editing tools and the emergence of deepfake technologies have made audio forgery a serious threat in various fields such as law, media, and cybersecurity. Ensuring the authenticity and integrity of audio recordings has become increasingly crucial. This paper presents a comprehensive digital audio forgery detection survey highlighting traditional feature-based and deep learning-based techniques. We introduce a novel classification of detection techniques and discuss various types of forgery operations including splicing, and copy-move. Additionally, we provide an overview of benchmark datasets, evaluation metrics, and open challenges in the field. Special emphasis is given to emerging technologies such as artificial intelligence-generated audio and blockchain-based integrity verification. This survey aims to guide researchers by identifying research gaps and suggesting promising directions for future work.