A Study Of Anonymous Communication Based On Voiceprint Deidentification

In the Internet age,deep learning technology is a new feasible solution to protect personal privacy.In public social networking,users' voice features and facial features lack specific privacy protection in terminals and links.The thesis combines speech recognition,speech synthesis,voiceprint recognition and signal processing,and proposes a secure communication scheme based on speech technology.The thesis scheme is divided into local side and server side.On the local side,the signal processing method blurs the content of the speech data by adding noise.On the basis of preserving the human auditory perception,the Mc Adams coefficient is used to suppress its voiceprint features;the model processing method is based on the deep speech learning model,combining speech recognition and Speech synthesis,using text as an intermediary,realizes the removal of voiceprint features.On the server side,the SFU scheme is used to build a communication service,and the WEB interface realizes the interaction and communication with the user.The contributions of the paper are as follows:(1)Propose a secure communication method based on the protection of identity information to help users obtain additional protection of identity privacy in communication and reduce the possibility of the terminal leaking personal voiceprint features;(2)Propose a security detection method,Based on the limited intelligibility and the anonymity of the voiceprint,combined with the word error rate of speech recognition,the cosine distance of voiceprint recognition,the delay of the communication network,and the perception of human hearing,multi-angle detection is implemented for the solution;(3)Realize a multi-mode secure communication application.According to different security requirements of users,three different identity privacy security protections can be used: default mode,fast mode,and strong security mode.