Deep Learning-Based Speech Detection for AI Clones and Ocean Acoustics Applications

Abstract This thesis is based on deep learning technology and proposes a new AI-cloned voice detection technique for practical application in police anti-AI precision fraud. The experimental validation results using the 2019 ASVspoof dataset show that the method has high detection accuracy and effectiveness. The technique is expected to prevent AI precision fraud actively, providing new ideas and strategies for research and application in related fields. This research aims to improve the security and accuracy of the voiceprint recognition system, which is essential for future research and application. Notably, similar deep learning methods have also been applied in ocean acoustics measurements, demonstrating the broad potential applications of this technology.