AI-Generated Voice Detection Using Machine Learning and Deep Learning

Abstract

Recent advancements in artificial intelligence have enabled the generation of highly realistic synthetic voices using text-to-speech and voice cloning technologies. While these innovations have numerous applications, they also introduce serious security threats such as impersonation, fraud, and misinformation. This paper proposes an offline AI-based system for detecting whether an audio sample is human or AI-generated. The system utilizes audio preprocessing techniques, feature extraction using LibROSA, and classification through machine learning and deep learning models including Random Forest and Convolutional Neural Networks. Additionally, sentiment analysis is incorporated to analyze emotional tone in user inputs, enhancing system intelligence. Experimental results demonstrate high classification accuracy and robustness, making the system suitable for forensic and cybersecurity applications.