Research On Encrypted Speech Biological Hash Retrieval Algorithm Based On Content Protection

Cloud storage technology as an important technology in cloud computing promotes the manufacture,dissemination and storage of multimedia data such as speech.However,the semi-openness of the cloud environment makes it difficult to ensure the security of speech data,which is very likely to lead to the leakage of users’ private information.Therefore,when sensitive data is stored in cloud servers,how to achieve efficient and accurate speech retrieval while protecting user privacy has become a current research hotspot in the field of speech retrieval.Based on the above problems,this thesis conducts an in-depth study based on content-based speech retrieval.The main research contents are as follows:1.In order to improve the robustness,discrimination and security of existing speech retrieval algorithms and to solve the problems of insufficient retrieval efficiency and accuracy in retrieval performance,an encrypted speech biometric hashing retrieval algorithm based on double hash index is proposed.Firstly,the spectral flux and kurtosis factor features of speech signal are extracted at the server,and then the two features are fused;Second,classify the speech signal and construct the key distribution index table based on the classification result;Then,according to the key distribution index table,the biometric security template with a single mapping key is established,and its biometric hash is quantized to obtain the hash index;At the same time,the mixed domain scrambling encryption is used to encrypt the original speech and construct the encrypted speech database;Finally,the hash index and encrypted speech library are uploaded to the cloud and building a cloud-based biohash index table.From the experimental analysis,it can be seen that the hash sequences generated after feature fusion well balance discrimination and robustness;meanwhile,the constructed double hash index table can achieve fast query and improve the retrieval performance of the algorithm.2.An encrypted speech biometric hashing retrieval algorithm based on audio segmentation is proposed,which can achieve the purpose of fast and accurate retrieval of target speech from large-scale speech data.The scheme consists of two phases: offline pre-processing and online retrieval.In the offline pre-processing phase,firstly,the PNCC features of the speech data are extracted and then feature classification is performed to construct a feature security template with a single mapping key.Then,slice the original speech into short-time audio segments according to the proposed audio segmentation algorithm,and the hash reconstruction operation is performed on the biohashing sequences to obtain the reconstructed hashing sequences according to the short-time audio segments.Finally,the original speech is encrypted using a nonlinear chaotic encryption algorithm to build a encrypted speech library.The online retrieval phase responds to the users’ query requests,just find the hash index that matches the query hash sequence from the biohashing index table.From the experimental analysis,it can be seen that the proposed audio segmentation and hash reconstruction technique can effectively remove the speech mute segments and optimize the data storage space by reducing the data volume;meanwhile,the classification of speech data can realize the second-level step-by-step retrieval,and the standardized edit distance is used for the hash sequence distance metric to realize the matching retrieval of different length sequences,which further improves the retrieval performance.This thesis is based on content-protected speech retrieval algorithm,combined with speech feature extraction,biometric template,audio segmentation,speech encryption and other techniques for in-depth research.The experimental results show that both proposed schemes have good retrieval performance and can achieve accurate and efficient secure speech retrieval.In future research,different types of malicious attacks will be further investigated to achieve tampering detection and recovery of speech after the attacks.