International Journal of Soft Computing and Engineering (IJSCE)
In the recent time, person authentication in security systems using biometric technologies has grown rapidly. The voice is a signal of infinite information. Digital signal processes such as feature extraction and feature matching are introduced to authenticate person in security system. In this paper concept of speaker recognition is discussed. Several methods such as Liner Predictive Coding (LPC), Mel Frequency Cepstral Coefficients (MFCCs), etc. are utilized for feature extraction and methods like Dynamic Time Warping (DTW), Vector Quantization (VQ), Hidden Markov Model (HMM), Gaussian Mixture Models (GMM), etc. are used with a view to identify voice signal.