Bu öğeden alıntı yapmak, öğeye bağlanmak için bu tanımlayıcıyı kullanınız: http://hdl.handle.net/11452/29102
Başlık: Investigation of the effect of data duration and speaker gender on text-independent speaker recognition
Yazarlar: Uludağ Üniversitesi/Mühendislik Fakültesi/Elektrik Elektronik Mühendisliği Bölümü.
Hanilçi, Cemal
Ertaş, Figen
S-4967-2016
AAH-4188-2021
35781455400
24724154500
Anahtar kelimeler: Computer science
Engineering
Support vector machines
Of-the-art
Verification
Adaptation
Classification (of information)
Face recognition
Support vector machines
Vector quantization
Linear discriminants
Recognition accuracy
Recognition performance
Speaker recognition
Speaker recognition evaluations
Speaker recognition system
Text independents
Training and testing
Speech recognition
Yayın Tarihi: Şub-2013
Yayıncı: Elsevier
Atıf: Hanilçi, C. ve Ertaş, F. (2013). "Investigation of the effect of data duration and speaker gender on text-independent speaker recognition". Computers and Electrical Engineering, 39(2), 441-452.
Özet: Duration of training/test data has a considerable effect on the performance of a speaker recognition system. In this paper, we analyze the effect of training and test data duration and speaker gender on the performance of speaker recognition systems. Gaussian mixture models-universal background model (GMM-UBM), vector quantization-universal background model (VQ-UBM), support vector machines generalized linear discriminant sequence kernel (SVM-GLDS) and support vector machines with GMM supervectors (GSV-SVM) are the classifiers we use for speaker recognition. Experimental results conducted on NIST 2002 and NIST 2005 speaker recognition evaluation (SRE) databases show that recognition performance breaks down when short utterances are used for training and testing independent from the recognizer (e.g. equal error rate (EER) reduces from 10.33% to 27.86% on NIST 2005) and GSV-SVM system yields higher EER than other methods in the case of using short utterances. It is also shown that recognition accuracy for male speakers are higher than female independent from database and classifier.
URI: https://doi.org/10.1016/j.compeleceng.2012.09.014
https://dl.acm.org/doi/10.1016/j.compeleceng.2012.09.014
http://hdl.handle.net/11452/29102
ISSN: 0045-7906
1879-0755
Koleksiyonlarda Görünür:Scopus
Web of Science

Bu öğenin dosyaları:
Bu öğeyle ilişkili dosya bulunmamaktadır.


DSpace'deki bütün öğeler, aksi belirtilmedikçe, tüm hakları saklı tutulmak şartıyla telif hakkı ile korunmaktadır.