Speech Encryption Technology Research Based On Embedded System

Information has become an important resource for the development of modern society with the process of social information.The popularization of the internet technology has accelerated the dissemination and the sharing of information resources.People pay more and more attention on how to guarantee the security of information in the open network.Speech signal is one of the most basic means of communication in our life.The application of voice communication in the whole society becoming more and more common,individuals and enterprises,countries need voice encrypted in some occasion,so that the research of speech encryption technology is very necessary.In reality,the speech encryption process usually has the voice signal compressed,and then have the compression voice data encrypted with the text encryption technology.The speech signal has the characteristics of high redundancy.With this text encryption technology it is hard to reduce the redundancy of speech signal,causing the encrypted data unsafe,so looking for the new encryption method is very important.In this topic author designed a new speech encryption method based on BSS.Blind source separation technology refers to extracting the original signal from the observation signals under the condition of the absence of the transmission channels parameters.For the reason that the blind source separation technology can complete separation of unknown transmission channel parameters,we can first mix multiple groups of speech data,to change the redundancy of speech signal.Then we can separate the mixed speech signal with blind source separation technology,the realization of speech signal encryption and decryption.For the reason of that the blind source mixing matrix can be randomly selected,so the processed speech signal and the original speech signal is not one-to-one correspondence,increasing the difficulty of cracking.Blind source separation have the intractability of underdetermined BBS,where the number of observed signals is less than source signal,signal cannot be separated,guaranteeing the security of the encrypted speech signal.According to that the signal with BBS have the uncertain problem of amplitude and order,extract the identification information of amplitude and order in each speech signal group before encrypting for voice restoration after BBS.In the process of BBS mixture there is pseudo random number generated by key-seed production,not only completing a cloak for speech signal but also introducing the key in the encryption process.There is a new underdetermined blind mixing matrix improved based on separability theory of mixture matrix,with that the encrypted signal cannot be group separated and have a good effect of cover.In decryption,considering that the BBS having large computation,the design of fast BBS algorithm is proposed to accelerate the speed of BBS.Finally,the article designs and debugs the whole hardware circuit of blind source voice encryption,constructing the SOPC system of speech encryption processing in FPGA.Then the article completes the software design of speech encryption in the embedded platform,analyzing and verifying the encryption effect and safety,drawing conclusion that the blind source encrypted voice signal is well and can't be decoded.