Research On Encrypted Speech Authentication And Recovery Algorithm Of Resisting Tampering Attacks

With the rapid development of the cloud industry,the application of cloud computing,cloud transmission and cloud storage are becoming more and more extensive,and the cloud platform has become the first choice for data transmission and storage with its many advantages.Ensuring the privacy security and integrity verification of multimedia data is particularly important in the field of data security.In this paper,improvement measures are proposed to address the issues of poor privacy and susceptibility to small-scale malicious tampering of the existing speech authentication system,and the integrity verification of user authentication information,small-scale tampering location and tampered content recovery are studied.The specific research tasks are as follows:1.A perceptual hash-secure speech authentication algorithm based on an improved chain DES in cloud environment is proposed to address the issues of speech feature and hash sequence leakage in authentication systems,as well as the detection and localization of authentication content after small-scale tampering attacks.Firstly,the NTEO and improved uniform subband frequency band variance features are extracted respectively,and the two are fused to construct feature vectors and binarized to generate hash sequences.Secondly,the hash index is generated by the improved chain DES algorithm to encrypt the hash sequence,and cloud encryption hash database is established.Then,a hash template is constructed to establish a one-to-one correspondence between the perceptual hash and the encrypted hash.Finally,the authenticated cloud encrypted hash value is matched with the hash to be authenticated after the hash template by Hamming distance algorithm.For the authenticated speech,the small-scale tampering attack and content-preserving operation are determined,and then the segment traversal algorithm and the difference method are used for small-scale detection and localization.The experimental results show that the proposed method can directly complete the authentication in the cryptographic hash,which not only effectively balances the discrimination and robustness,but also can accurately locate the malicious substitution and muting attack regions.In addition,the proposed encryption algorithm introduces a random mechanism of key control to encrypt the hash sequence in a “one-at-a-time” manner,and to remove the correlation between the hash and the input,improving the security of the hash in the authentication system.2.To address the issues of inaccurate identification of small-scale tampering attacks in existing speech hash authentication algorithms,inaccurate localization in ciphertext speech,and inability to achieve content recovery after continuous tampering,a ciphertext speech Bio Hashing authentication algorithm that can resist tampering attacks is proposed for speech content authentication and tamper recovery in cloud-based encrypted domains.Firstly,the chaotic key and scramb-shift algorithm were combined to construct a cloud ciphertext speech database,and Gammatone Frequency Cepstral Coefficients(GFCC)were extracted based on human auditory perception model to construct robust feature vectors.Then,the orthogonal matrix was constructed after Schmidt-orthogonality by two-dimensional Logistic-adjusted Sine(2D-LASM)mapping,and the dimension vector and feature vector of the matrix are used as internal product to construct a biosecurity template,and the Bio Hashing sequence corresponding to the ciphertext speech is further quantified and generated,uploaded to the cloud biological hash database for storage.Finally,the small range of authenticated tampered speech combined with frame information is tampered detected and located,and the tampered samples are recovered by linear prediction coefficients.The experiment shows that this algorithm has the ability to locate small-scale tampering,and it can accurately locate the regions tampered by malicious replacement and muting attacks without embedding detection information.Similarly,the tampered samples can be recovered without any auxiliary watermark information.The constructed biosecure template has high security and complexity,and the Bio Hashing and ciphertext speech can effectively prevent the leakage of private data,with high security and privacy.