Study and Simulation of the Acoustic Features in Speaker Recognition System

In speaker recognition system, it抯 one of the key problems to extract the valid acoustic features that can represent speaker抯 characters. Several kinds of important features are studied and extracted, include linear prediction cepstrum coefficients (LPCC), mel-frequency cepstrum coefficients (MFCC), acoustic dynamic feature etc. These features are analyzed and compared, and the mixed features?effect on the performance of the recognition system is also researched. The simulation results show that the mixed features can obviously improve the recognition accuracy of the speaker recognition system.