Secure Voice Processing Systems against Malicious Voice Attacks

This book provides readers with the basic understanding  regarding the threats to the voice processing systems, the  state-of-the-art defense methods as well as the current research results  on securing voice processing systems.It also introduces three  mechanisms to secure the voice processing systems against malicious  voice attacks under different scenarios, by utilizing time-domain signal  waves, frequency-domain spectrum features, and acoustic physical  attributes.

First, the authors uncover the  modulated replay attack, which uses an inverse filter to compensate for  the spectrum distortion caused by the replay attacks to bypass the  existing spectrum-based defenses. The authors also provide an effective  defense method that utilizes both the time-domain artifacts and  frequency-domain distortion to detect the modulated replay attacks.  Second, the book introduces a secure automatic speech recognition system  for driverless car to defeat adversarial voice command attacks launched  from car loudspeakers, smartphones, and passengers. Third, it provides  an acoustic compensation system design to reduce the effects from the  spectrum reduction attacks, by the audio spectrum compensation and  acoustic propagation principle. Finally, the authors conclude with their  research effort on defeating the malicious voice attacks and provide  insights into more secure voice processing systems.

This  book is intended for security researchers, computer scientists, and  electrical engineers who are interested in the research areas of  biometrics, speech signal processing, IoT security, and audio security.  Advanced-level students who are studying these topics will benefit from  this book as well.