Voice authentication module using mel-cepstral coefficients

Abstract

Objective. The purpose of the study is to develop and apply a method for extracting information about the identity of users from recordings of their voices using the calculation of mel-cepstral coefficients.Method. In the study of the application of methods for extracting informative features from a voice recording, allowing identification of the speaker, an authentication scheme using mel-cepstral coefficients is presented.Result. Based on this method, an authentication module was implemented using audio recordings of user voices using the simplest MFCC. The authentication module was developed using Python languageConclusion. The biometric authentication method is an inexpensive and relatively simple way to verify the authenticity of users. Despite the obvious advantages of mel-cepstral coefficients, this method has certain disadvantages. To eliminate shortcomings, various frequency filters can be used, as well as third-party algorithms for analyzing audio recordings.