Biometric Authentication System for Industrial Applications using Speaker Recognition

Biometric authentication has gained popularity in recent years as knowledge-based authentication methods overburden users. This is because users are required to remember distinct and secure passwords for each system where they are registered. As speech recognition systems are embraced the smart technologies of the future, speaker recognition is going to be a natural authentication mechanism. This paper presents a biometric authentication system for industrial applications using speaker recognition. The system uses an autocorrelation voice activity detector and gaussian mixture models to identify users of the system. The equal error rate of the overall designed system was found to be 12.1% and the average verification time was found to be 4.67 seconds.