Publication: Arabic Speaker Identification System for Forensic Authentication Using K-NN Algorithm
Date
2021
Authors
Abdulwahid S.
Mahmoud M.A.
Abdulwahid N.
Journal Title
Journal ISSN
Volume Title
Publisher
Springer Science and Business Media Deutschland GmbH
Abstract
For many years, there was an increasing necessity for being capable of the identification of a person based on his/her voice. Judges, law enforcement agencies, detectives, and lawyers, wanted to be able to use the forensic authentication of voice for investigating a suspect or confirming a judgment of guilt or innocence. This study aims to design and build a comprehensive identification of the forensic speakers for the Arabic language. The suggested system has been utilized for the recognition of forensic speaker�s isolated words for purposes of identification. It comprises two stages; the first stage is training the sentence of the forensic speaker in the case where it is not previously processed and stored; the second stage is testing; it is applied in the case where the sentence of the forensic speaker has been previously processed and stored. Every one of the phases involves utilizing audio features (standard division, mean, amplitude, and zero-crossing), pre-processing with the use of the MFCC, Hamming Window, vector quantization, and data mining classification approaches. The proposed system implementation provides removal of the noise in spoken sentences, processing speech sentences prior to the storing, and a correct classification with the use of a number of algorithms of data mining classification such as the Logistic Model Tree (LMT), and K-nearest neighbor (KNN) algorithms. KNN being given the highest accuracy of 91.53% and 94.56% respectively. � 2021, Springer Nature Switzerland AG.
Description
Classification (of information); Data mining; Digital forensics; Forestry; Learning algorithms; Loudspeakers; Motion compensation; Nearest neighbor search; Speech recognition; Trees (mathematics); K-near neighbor; Logistic model tree; Logistics model; Mel frequency cepstral co-efficient; Mel frequency cepstral coefficient; Mel-frequency cepstral coefficients; Mining classification; Model trees; Nearest-neighbour; Speaker identification systems; Authentication