文章基本信息

标题：A Study of Speaker Recognition Approaches Based on Feature Selection and Classification Methods
本地全文：下载
作者：Hasti Baharipour ; Mohammad EbrahimShiri Ahmadabadi ; Mohammad Mosleh 等
期刊名称：International Journal of Computer Science and Network Solutions
印刷版ISSN：2345-3397
出版年度：2014
卷号：2
期号：9
页码：61-67
出版社：International Journal of Computer Science and Network Solutions
摘要：One of the main ways of controlling accesses and developing security in cybernetics in order to protectinformation is recognizing and confirming individuals’ identity. Compared with other biometric methods,recognizing people through their voices is economical since it needs no special expensive equipment.Classifier, as one main part of speaker recognition system, has an important role in improving its function.Extracting speech features is one of the most important topics in the field of recognizing and identifying aspeaker. The main purpose of extracting features is removing unnecessary information from speech signaland changing the speech signal into a format which makes classes separation easy in pattern recognitionphase. However, all extracted features are not useful or effective. Nowadays, different methods are used forspeaker recognition. In this article, we have explain two methods of speaker recognition method based onmodeling methods and speaker recognition method based on feature selection. We also review theconclusion of implementing these methods
关键词：Speaker Identification; Speaker Verification; Classification; feature Selection; Gaussian;MixtureModel (GMM); Support Vector Machine (SVM); Genetic.