Research On Voiceprint Authentication For User Privacy Protection

The application of the informational property management system in the real estate industry has made it easier for head of households to enjoy property services.When using the system,the certification of users' identities is an indispensable part.The traditional text-based password authentication method has the problems of being forgotten and stolen easily.In contrast,the authentication method based on voiceprint features not only overcomes the above drawbacks,but also has the advantages of high user acceptance,low acquisition equipment cost,convenient remote authentication and so on.Therefore,it is of great significance to design and implement an authentication module based on the user voiceprint for the informational property management system on the mobile terminal.However,the user's original voiceprint feature information is unique and immutable.It belongs to the user's privacy data.Once intercepted by the attacker,it will pose a long-term threat to user information security.Therefore,researching and designing the voiceprint identity authentication scheme that can protect the user's privacy has important practical significance and research value to improve the security of the informational property management system identity authentication.This paper studies the voiceprint authentication method for user privacy protection for informational property management systems.The main research contents are as follows:(1)Optimization of voiceprint features.In order to solve the problem that the voice collected during the authentication process often has noise interference,the PNCC feature which has stronger noise robustness is used as the feature parameter of voiceprint authentication,and its extraction process is optimized.(2)Voiceprint authentication algorithm design for user privacy protection.A voiceprint authentication algorithm for user privacy protection based on homomorphic encryption and vector quantization is designed.In the registration algorithm,the public key of Paillier homomorphic encryption is used to perform homomorphic encryption on the user's voiceprint feature codebook.In the authentication algorithm,for the problem of low encryption efficiency of homomorphic encryption algorithm,AP clustering algorithm is used to cluster the PNCC features collected by the client,and then homomorphic encryption is performed on the clustered features.Then,the Euclidean distance component difference between the clustered features to be authenticated and the feature codebook is calculated in the ciphertext domain to avoid the direct exposure of the original voiceprint feature on the server side and achieve the identity authentication while protecting the privacy of the user.Simulation experiments show that the improved voiceprint authentication algorithm can not only maintain the recognition rate,but also significantly reduce the encryption time and improve the realtime nature of the authentication process.(3)Voiceprint authentication protocol design.In order to enhance the security and privacy of the voiceprint template in the transmission process and its ability to defend against attacks,the registration protocol,authentication protocol,template update protocol,and template cancellation protocol are designed to combine SSL two-way authentication technology with homomorphic encrypted voiceprints.The combination of authentication algorithms ensures the secure transmission of the user's voiceprint features in the channel.(4)The realization of voiceprint identity authentication for user privacy protection in the informational property management system.Using the C/S architecture,utilizing the convenience of collecting voice signals by the mobile phone terminals,the development of Android mobile voiceprint authentication client and the development of PC-side voiceprint authentication server.The performance test was performed on the system.The test results showed that the scheme designed by the dissertation could achieve the function of logging in the informational property management system through the voiceprint while the recognition rate reached the system demand.