Fulltext

Text-Dependent Audio Biometric Person Authentication

Ekhlas Falih Naseer *

IRAQI JOURNAL OF COMPUTERS,COMMUNICATION AND CONTROL & SYSTEMS ENGINEERING المجلة العراقية لهندسة الحاسبات والاتصالات والسيطرة والنظم
ISSN: 18119212 Year: 2012 Volume: 12 Issue: 2 Pages: 56-68
Publisher: University of Technology الجامعة التكنولوجية

Abstract

Abstract: Audio biometric person authentication is the task of verifying the identity of a person based on the information in the speech signal that occurs during the production of speech. In this research, audio person authentication is focus on acoustic text-dependent speaker verification system. The proposed recognition process begins by converting that audio into frequency domain by applying discrete cosine transform (DCT), then compute the seven moments as a features for that audio and build a database depends on these features, then compute the Kohonen neural network for person identification, and then compute the dynamic time wrapping (DTW) for verification and patterns matching, and later give the decision logic for accepting or rejecting a claimant.

Keywords

Audio Biometric Person Authentication --- Multi modal Speech Processing --- Discrete Cosine Transform --- Rules of Moments --- Kohonen Network Rules --- Dynamic Time Wrapping.