In biometrics, identification of an individual human being is based on the speech signal generated by their speech production system. In this paper, a novel text independent speaker recognition system using the physiological characteristics of human speech production system is developed. Pitch, fundamental period and the number of peaks are the features considered for the development of speaker recognition system. Feature extraction based on Auto-correlation, Discrete Wavelet Transform and Cepstrum techniques are performed. Evaluation of the system performance for the various feature extraction techniques is done in terms ofsensitivity.