A Review of Automatic Speaker Verification Systems with Feature Extractions and Spoofing Attacks

Automatic Speaker Verification (ASV) technology is gaining traction in various applications due to its reliance on a unique and natural biometric identifier: a person's voice. This review provides a comprehensive analysis of ASV systems, exploring their core functionalities, underlying principles, and potential benefits. It examines the enrollment and verification processes, where speaker voiceprints are created and compared for user authentication. The review then delves into feature extraction techniques, crucial for differentiating genuine and spoofed voices. Common spoofing attacks like replay attacks and voice mimicry are discussed, along with feature analysis methods for spoof detection. Achieving high accuracy, precision, reliability and security in Automatic Speaker Verification (ASV) systems, particularly in the presence of spoofing attacks, involves a combination of robust feature extraction techniques, advanced machine learning models, and effective anti-spoofing strategies ASVSpoof dataset is used for this research. The study concludes by highlighting ongoing challenges and future directions in ASV research, emphasizing the importance of security, fairness, and continuous advancements in this evolving field.