Research On Identity Authentication Mechanism Based On Speech Feature

Currently, authentication is one of the most popular topic as the first checkpoint in the network security. The traditional authentication is based on password. However, the server is hard to judge the authenticity of a user’s identity if the password is stolen. So traditional authentication need to improve, to increase their security. The speech feature has the advantages of uniqueness, security, stability, reliability, and so on, which make it to be used widely in the authentication. According to these advantages of speech features, the author put forward the identity authentication mechanism based on speech feature on the basis of the traditional identity authentication, add to the speech features as speech verification code of traditional identity authentication, which increases the security of authentication.This paper has completed the analysis of speech signal and extracted the speech features LPC, LPCC and MFCC by the process of the pre-emphasis, frame and adding window, endpoint detection, feature extraction and so on for the speech signal. The paper devote to solve two difficult problems of the integrity and non-repudiation of the identity authentication by the combine speech features and SHA-1 algorithm to complete message authentication of the user; The paper solved the two problems of the credibility and access control in the identity authentication by the DTW speech recognition algorithm and response -challenge mechanism of the combination of speech verification code. In the end, under the condition of VC6.0, the paper completed the test of the SHA-1 message authentication algorithm based on speech features. Input is"0-9" ten digital speeches, the hashed value of the output is completely different. The conclusion of considering the speech feature as identifying of message authentication is very safe and reliable. And the paper has finished the test of the DTW (Dynamic Time Warping) speech recognition algorithm by platform of MATLAB, that is to test the recognition rate of "0-9" ten digital speeches in the different environment, which shows that the recognition rate of the DTW algorithm for "0-9" ten digital is very high. So this paper combined challenge-response mechanism with"0-9", which put forward to random select six digital from the ten digital as the speech verification code of the identity authentication.