Gender recognition of human based on speech characteristics by features fusion with K_NN and MLPNN classifications

The desire to establish human-computer speech communication, as well as the expansion of the use of the Internet and the possibility of receiving various information services and remote services, has caused the increasing importance of various speech processing techniques. In fact, people’s speech p...

Full description

Saved in:
Bibliographic Details
Main Authors: AL-Dujaili, Mohammed Jawad, Ahily, Hydr Jabar Sabat, Fatlawi, Ahmed
Format: Conference Proceeding
Language:English
Subjects:
Accuracy
Artificial intelligence
Classification
Coefficients
Gender
Information services
Linear prediction
Principal components analysis
Speech
Speech processing
Speech recognition
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The desire to establish human-computer speech communication, as well as the expansion of the use of the Internet and the possibility of receiving various information services and remote services, has caused the increasing importance of various speech processing techniques. In fact, people’s speech plays an important role in analyzing their personality. You can tell the gender of people from their voices. Currently, the automatic detection of human gender by machines has received special attention in the field of artificial intelligence. Speech recognition is a subjective phenomenon and, in this field, many researches have been conducted in recent years and different methods have been presented. Features of Capstral coefficients based on MFCC Mel and Capstral coefficients based on LPCC and LPC linear prediction are three successful features in the field of speech classification, especially in detecting the speaker’s gender. In this article, a new method has been presented by using the features of MFCC, LPCC, and LPC and combining them with KNN and MLPNN classification, which increase the classification accuracy. The useful coefficients in these three features are combined and used to recognize the gender of the speaker. Before using the categories and to reduce the dimensions of the feature vector, we used the principal components of PCA. A TIMIT dataset has been used for this purpose. The presented method provides a high accuracy of 97% in detecting the male gender and an accuracy of nearly 98% in detecting the female gender.
ISSN:0094-243X
1551-7616
DOI:10.1063/5.0181969