A robust content authentication algorithm of speech based on perceptual hashing

作者：Dong Rui Hong; Di Yan Jun; Zhang Qiu Yu

来源：Sensors and Transducers, 2013, 161(12): 375-382.

• Authentication algorithm
• Communication terminal
• Content authentication
• Line spectrum pair
• Linear prediction analysis
• Perceptual hashing
• Speech content
• Tamper localization

To solve the problem of real-time authenticating in traditional speech authentication algorithms, an efficient speech content authentication algorithm of perceptual hashing is proposed. Firstly, the speech signal is preprocessed, linear prediction analysis is performed on each frame to get its 10-th order LPCs and then transform them into line spectrum pair (LSP) coefficients by Discrete Fourier Transform(DFT) as perception characteristics; Secondly, speech data are divided into groups orderly, combinative the weighted expectations of each group' LSP coefficients to be the authentication data, and the quantity of authentication data are compressed through hashing construction; Finally it realize the fast authentication through hashing matching. Experiment results show that the algorithm is robust to content preserving operations, such as changing volume, echo, resampling, etc. And it is fragile to malicious operations including replacing, deleting, etc, and also with the ability of locating the tampered position precisely. It has the properties of small authentication data and high running efficiency and applies to limited-sources communication terminals.

• 出版日期2013