A Forensic Speaker Recognition System for the Arabic Language

Nassren N. Abd Alwahed and Dr. Talib M. Jawad

For a long time, there has been a need to be capable of identifying an individual based on their voice. For a number of years, law enforcement agencies judges, lawyers, and detectives, wanted to utilize forensic voice authentication for investigating a suspect or confirming a judgment of innocence or guilt. The paper aims to design and build an efficient forensic speaker identification for Arabic language. The proposed system is used to recognize the forensic speaker sentences for the identification purpose. It includes 2 phases; The 1st one is the training of the forensic speaker sentence when it is not processed and stored previously; The second phase is the testing phase which is implemented if the forensic speaker sentence is stored and processed previously. Each phase of the phases includes the using of audio features (mean, standard division, zero crossing, amplitude), preprocessing using Hamming Window, MFCC, vector quantization and data mining classification algorithms. The implementation of the proposed system provides noise disposal of the spoken sentences, processing the speech sentences before storing, and an accurate classification using some of the data mining classification algorithms like KNN, SMO, and NB.

Volume 11 | 04-Special Issue

Pages: 1-7