USING THE WINDOW FOURIER TRANSFORM IN VOICE BIOMETRIC IDENTIFICATION APPLICATIONS

Abstract

Currently, biometric identification systems are an integral part of our life. Fingerprint 
scanners built into smartphones, voice or speech recognition technologies and other tools 
are gradually replacing the traditional methods of identification and increasingly penetrate 
into such large areas as banking, visa and migration services, medicine, security systems, 
etc. Biometric systems have a number of advantages over traditional methods, since they 
are adapted for personal identification without the possibility of transferring a secret key, 
and are in many ways more convenient from the user's point of view. However, the more 
actively the implementation of this type of systems is, the more acute the issue of ensuring 
information security arises. 
With the intensive development of biometric identification and authentication tech
nologies in the world, there is an increased interest in the development of innovative solu
tions for voice biometrics. The voice is as much an integral part of every person as their 
face or fingerprints. Widespread communication means open up great opportunities for the 
use of this identifier. In addition, voice recognition is very convenient for users and requires 
minimal effort from them. Voice authentication systems rely on voice characteristics 
unique to each person, such as pitch, modulation and frequency of sound, during sample 
recording and in the process of subsequent identification. These indices are determined by 
the physical characteristics of the vocal tract and are unique to each person. 
A technique for analyzing a voice signal using a short-term Fourier transform with 
different window functions is introduced in the present work. The main limitations of the 
traditional methods of Fourier transform for voice identification problems are considered. A 
voice recognition module is modeled in the LabVIEW 2020 software space.