Performance of Speaker Recognition
System Using Kernel Functions
Approach for Different Noise Levels
Renu Singh, Arvind Kumar Singh, and Utpal Bhattacharjee
Abstract Speaker recognition is one of the most popular voices biometric tech-
nique used for security reason in many areas like banking system, online-shopping,
database access etc. The recognition performance of speaker recognition system is
very satisfactory in noise-free environments, whereas the improved performance in
case of low level signal-to-noise ratio (SNR) is the need in the present days. Hence,
in this study speaker recognition performance has been evaluated at different signal-
to-noise ratios using SVM-based various kernel functions approach and principal
component analysis (PCA). The proposed scheme has been applied on NIST 2003
and AURORA dataset and found that the recognition accuracy and running time
improves at low level SNRs using kernel functions.
Keywords Speaker recognition system · SVM · Principle component analysis
(PCA) · Signal-to-noise ratio
1 Introduction
Speaker recognition is a method of recognizing a speaker’s voice by utilizing explicit
information comprised in speech waves [1, 2].This methodology can be used to
substantiate the characteristics claimed by the person who wants to access the systems
i.e. enabling the access control of different services with the help of voice. The various
applications of speaker recognition system are voice dialing, phone banking, shop-
ping over phone, database access services, reservation services, services on voice mail
R. Singh (B ) · A. K. Singh
Department of Electrical Engineering, North Eastern Regional Institute of Science and
Technology, Nirjuli, Arunachal Pradesh 791109, India
e-mail: renumona08@gmail.com
A. K. Singh
e-mail: singharvindk67@gmail.com
U. Bhattacharjee
Department of Computer Science, Rajiv Gandhi University, Itanagar, Arunachal Pradesh 791112,
India
e-mail: utpalbhattacharjee@rediffmail.com
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer
Nature Singapore Pte Ltd. 2020
P. K. Mallick et al. (eds.), Electronic Systems and Intelligent Computing, Lecture Notes
in Electrical Engineering 686, https://doi.org/10.1007/978-981-15-7031-5_49
513