Research and comparison of deepfake audio detection algorithms

• Main Author: Mo, Fei
• Other Authors: Alex Chichung Kot
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
• Online Access: https://hdl.handle.net/10356/158873
• Institution: Nanyang Technological University

Similar to other biometric systems, speaker verification systems are easy to be affected by various spoofing attacks. In recent years, there have been more and more researches on deep learning, and many important advances have been made, artificially synthesized pronunciations are getting closer and closer to real human speech. This progress has made important contributions to many fields such as voice navigation systems and human-computer interaction, but also brought important security risks. Therefore, how to efficiently and accurately identify deepfake audio is very important. The main research work of this dissertation is as follows: (1) The basic process of deepfake audio detection is summarized, including preprocessing, feature extraction, classification detection (2) Two traditional models and three deep learning models are reproduced and the results are compared.