Significance of nature of features from linear prediction residual for speaker recognition using gaussian mixture model and ergodic hidden markov model
