MU Digital Repository
Logo

A Novel framework for Voice Signal Recognition using RASTA & Authentication by Neural Networks

Arjunan, Vijaya R and Kumar, Vijaya V (2009) A Novel framework for Voice Signal Recognition using RASTA & Authentication by Neural Networks. International Journal of Recent Trends in Engineering, 2 (3). pp. 82-84. ISSN 1797-9617

[img] PDF
ijrte02038284_voice_signal_recognition_RASTA.pdf - Published Version

Download (324kB)

Abstract

This Paper work presents a novel voice verification system using continuous wavelet transforms [1], [2] and additionally an improved authentication system of received voice signal through Back Propagation Neural Networks is implemented in addition to existing voice verification system [3]. Voice signals are comparatively more dynamic when compared to other analog signals. The conventional signal processing techniques are better only for stationary signals and non optimal for few dynamic signals. The proposed voice recognition system uses combination of the Relative Spectral Algorithm and linear prediction. The received voice signal is implemented for authentication using Neural Networks training algorithm. The received voice signal is first filtered using the special purpose voice signal filter called Relative Spectral Algorithm. The signal is modeled as a linear combination of its past values and present and past values of a hypothetical input to a system. The signals are denoised and decomposed further to derive the wavelet coefficients and thereby a statistical computation is carried out. Further the resonance of the voices signal is detected using the Neural Network which is used for authentication purpose. The performance of the proposed system is evaluated by using noisy speech signals. Test results show the effectiveness of the proposed speech recognition system. The rate of correct classification is about 93% for the sample speech signals.

Item Type: Article
Additional Information: © 2009 ACADEMY PUBLISHER
Uncontrolled Keywords: CWT - Continuous Wavelet transform, RASTARelative spectral algorithm, LPC- Linear predictive coding, BPN- Back Propagation Networks, MLFFN- Multi-layer feedforward networks, NN-Neural Networks
Subjects: Engineering > MIT Manipal > Computer Science and Engineering
Depositing User: MIT Library
Date Deposited: 28 Jul 2011 07:52
Last Modified: 28 Jul 2011 07:52
URI: http://eprints.manipal.edu/id/eprint/1011

Actions (login required)

View Item View Item