Open AccessSpeech biohashing authentication based on power spectrum
Author(s) -
Yibo Huang,
Tengfei Chen,
Xiangrong Pu,
Qiu-yu Zhang
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2010/1/012058
Subject(s) - computer science , hash function , pattern recognition (psychology) , feature vector , robustness (evolution) , biometrics , feature extraction , speech recognition , authentication (law) , artificial intelligence , algorithm , biochemistry , chemistry , computer security , gene
In order to solve hashing sequences leakage in the existing speech authentication algorithms, this paper proposed a speech biohashing authentication based on power spectrum. Fristly, the algorithm pre-processed the original speech, then used the improved covariance method to extract the power spectrum feature of the speech, and generated a one-dimensional feature vector from the feature matrix. A pseudo-random matrix was generated by using Sinc function transform, the feature vector and pseudo-random matrix are projected and mapped to construct the biometric template. Finally, binarization process was used to generate hash index, and upload it to the cloud. The experimental results show that the algorithm has good discrimination, robustness and authentication performance. At the same time, the biometric template has good security and ensures the irreversibility of the hashing sequence.