Research On High-efficiency Speech Perceptual Hashing Authentication Algorithm Based On Instant Speech Communication

With the rapid development of network communication and mobile communication technology,Speech has become one of the main ways for people to communicate instantly because of its advantages of the convenience of collection,natural and technical maturity.However,in the process of speech instant communication,the speech was easily affected by codec,channel noise,delay,and packet loss,and due to the openness of wireless network channel,it will be opening a door for the illegal attacks of hacking,malicious tampering and other.In order to achieve the efficient speech authentication solve the interactive effects of speech perception hash in robustness,discrimination,authentication efficiency and realize the tamper localization of speech,this thesis mainly research on the efficient speech perception hash authentication algorithm of instant speech communication,and under the condition of high speed and low reli ability of instant communication,an efficient speech perception hash algorithm is obtained which include noise reduction,tamper detection and localization.The main research contents are as follows:1.Aiming at the high efficiency requirement of percep tual hashing algorithm for instant speech communication,it is necessary that the perceptual hashing algorithm under the premise of content preserving operation robustness and different speech distinguishability have characteristics of small authentication data and high operating efficiency.Thus,an efficient speech perceptual hash authentication algorithm is presented base on the linear prediction-minimum mean squared error(LP-MMSE),this algorithm has good perceptual robustness to the content preserving operation and distinctiveness for different speech.Meantime,it also has other advantages,such as the characteristics of small abstract and high operating efficiency.Moreover,in order to solve the balance of the content preserving operation between th e perceptual robustness,distinguishability of different speech and authentication efficiency,an efficient speech perceptual hash authentication algorithm is applied base on the improved spectral entropy method.Through the experiments which use the Teager energy operator respectively combine the linear prediction coefficient(LPC),LP-MMSE and linear spectrum pair(LSP)coefficient comparing with each other shown that the proposed algorithm has good balance between the robustness,distinguishability and authentication efficiency,and satisfy the requirements of small abstract of perceptual hash and real time authentication of speech communication.2.Aiming at the perceptual robustness of content preserving operations and discrimination of speech content after the format conversion of heterogeneous mobile terminal.And in order to realize the requirement of noise reduction,tamper localization in small range and efficient certification,an efficient speech perceptual hash authentication algorithm is present ed based on spectral subtraction and energy entropy ratio method.This algorithm firstly used pretreatment to process the waiting of certified speech,and applied the spectral subtraction to reduce the noise.Then through calculating the speech which has p rocessed by the noise reduction,the parameter matrix of energy entropy of each frame using the energy entropy ratio method is obtained.Finally,the hash structure is used to generate the binary perceptual hash sequence.The experimental result shows that the algorithm can effectively reduce the noise of speech,and has good perceptual robustness of content preserving operations and discrimination of different speeches,and the according perceptual hash has the characteristics of small abstract,high authe ntication efficiency and able to achieves the tamper detection and localization in a small range.