Research On Retrieval Method For Encrypted Speech Based On Biological Hashing

Mass-based encrypted speech retrieval technology has broad research and application space,but the increase of cloud ncrypted speech data enables the existing search method to check the recall rate,precision rate,anti-noise ability and retrieval efficiency in mass application,encryption performance and other inefficiencies.Therefore,how to ensure the privacy of speech data,how to improve the retrieval efficiency and retrieval accuracy of encrypted speech data has very important research significance.In order to solve the problem of efficient matching retrieval of massive encrypted speech,the thesis uses encrypted speech as the research carrier,and comprehensively uses the theory and methods of speech signal processing,biological hashing,speech encryption,indexing,encrypted speech retrieval to search content-based ciphertext.The key technologies were analyzed theoretically and experimentally.The main research work is as follows:1.In order to meet the requirements of efficient extraction of speech features,good compactness and high security in speech perceptual hashing construction,a safety and efficient speech biological hashing algorithm is proposed.The algorithm generates a binary hash sequence based on the DWT and Mersenne Twister algorithms.The experimental results show that the proposed algorithm can quickly extract the speech biological hashing features,and it has good robustness and discrimination to the speech content maintenance operation,and has high security.2.In order to improve the robustness and discrimination of speech perceptual hashing function in content-based encrypted speech retrieval,improve retrieval efficiency and accuracy,and realize privacy security of speech data,a content retrieval algorithm of encrypted speech based on biological hashing was proposed.The algorithm firstly processes the speech using the two-dimensional Henon mapping encryption algorithm and uploads it to the cloud encrypted speech library.Then it combines DWT transform,Logistic map and FFT transform to generate binary hash sequence and store it in the cloud system hash index table.The Hamming distance matching search is used for the search.It can be seen from the experimental results that the hash function of this algorithm is very robust and distinguishable,and the retrieval efficiency,accuracy and security of the retrieval model are significantly improved.3.In order to improve the impact of noise on the robustness and discrimination of speech hashing schemes,improve retrieval efficiency,accuracy,and security of retrieval compactness,as well as integrity authentication of query results,a high efficiency encrypted speech retrieval scheme based on biological hashing and spectral subtraction is proposed.Firstly,the two-dimensional Arnold mapping encryption algorithm is used to encrypt the speech object and upload it to the cloud encrypted speech library.Then,the pre-processed speech signal is spectrally subtracted,and then the waveletautocorrelation function is used for the denoised speech signal.The chebychew map and the FFT transform generate a binary hash sequence and store it in the cloud system hash index table.When the user searches,the Hamming distance is used for matching search,and the search result is authenticated and matched.The experimental results show that the proposed scheme obviously deals with the noise of speech,has strong robustness and discrimination,and the retrieval efficiency,accuracy and security are obviously improved.