Study of Machine Learning Methods for Voice Biometric Identification in Data Protection Systems

EasyChair Preprint 15577

Abstract

This study analyzes and evaluates various machine learning methods applied to voice biometric identification in data protection systems. To reduce the impact of errors associated with voice biometric data during the preparation of models for training and testing, automated approaches and algorithms were developed. These algorithms include preprocessing of speech signals, automated segmentation, and extraction of voice and speech biometric features, as well as methods for multi-algorithmic and multimodal blending and assessment of speech material quality. Based on the prepared data, various machine learning models were trained, such as the Gaussian Naive Bayes classifier, Support Vector Machine (SVM), Perceptron, and k-Nearest Neighbors. After training, the accuracy of each model was evaluated.

Keyphrases: Cybersecurity, biometric identification, data protection, machine learning

@booklet{EasyChair:15577,
  author    = {Serhii Semenov and Viacheslav Davydov and Denys Grynov},
  title     = {Study of Machine Learning Methods for Voice Biometric Identification in Data Protection Systems},
  howpublished = {EasyChair Preprint 15577},
  year      = {EasyChair, 2024}}