Survey Maps AI Voice Generation and Detection Challenges

A new survey published on arXiv provides a comprehensive overview of AI-generated voices and their detection, highlighting the rapid advancement of AI models in creating realistic human voices and the associated risks. The study covers both the technical foundations and state-of-the-art methods for voice generation and detection, addressing unique challenges posed by phonetics, prosody, and auditory perception. It also identifies key open challenges, benchmark resources, and future research directions. The survey underscores the urgent need for robust safeguards against voice cloning scams, which have recently targeted businesses and political leaders, and notes the distinct difficulties in detecting synthetic voices compared to image and video deepfakes. The paper is categorized under Computer Science and Artificial Intelligence, with the abstract submitted on arXiv (ID: 2608.15411). The survey aims to be a valuable resource for future researchers in this field.