Voice Biometric Identity Authentication System Based on Android Smart Phone

This thesis studies and implements voice biometric authentication system on Android smart phone platform. Firstly, the characteristics of voice database and voice recognition and authentication system flow are introduced. Then, a random shuffling algorithm is used to generate a voice cipher. After this, the voice feature extraction algorithm based on MFCC is introduced. In this paper, GMM is introduced to train the voiceprint feature model of voice segment data effectively. When number of training samples is 5, the success rate of system authentication is between 89% and 96% and time required to authenticate a voice sample is between 210ms~320ms, which demonstrate the high accuracy and good real-time.