A new approach for Persian speech Recognition

Author

Pour, Meysam Mohamad ; Farokhi, Fardad

Author_Institution

Dept. of Electr. Eng., Centeral Azad Univ., Tehran

fYear

2009

fDate

6-7 March 2009

Firstpage

153

Lastpage

158

Abstract

In this paper in consideration of each available techniques deficiencies for speech recognition, an advanced method is presented that´s able to classify speech signals with the high accuracy (98%) at the minimum time. In the presented method, first, the recorded signal is preprocessed that this section includes denoising with Mels Frequency Cepstral Analysis and feature extraction using discrete wavelet transform (DWT) coefficients; Then these features are fed to multilayer perceptron (MLP) network for classification. Finally, after training of neural network effective features are selected with UTA algorithm.

Keywords

cepstral analysis; discrete wavelet transforms; feature extraction; learning (artificial intelligence); multilayer perceptrons; signal classification; signal denoising; speech recognition; Persian speech recognition; UTA algorithm; discrete wavelet transform coefficient; feature extraction; mels frequency cepstral analysis; multilayer perceptron network; neural network training; speech signal classification; Automatic speech recognition; Cepstral analysis; Discrete wavelet transforms; Feature extraction; Frequency; Hidden Markov models; Multilayer perceptrons; Neural networks; Speech recognition; Wavelet analysis; Discrete Wavelet Transform (DWT); Mels Scale Frequency Filter; Multilayer perceptron (MLP) neural network; UTA algorithm;

fLanguage

English

Publisher

ieee

Conference_Titel

Advance Computing Conference, 2009. IACC 2009. IEEE International

Conference_Location

Patiala

Print_ISBN

978-1-4244-2927-1

Electronic_ISBN

978-1-4244-2928-8

Type

conf

DOI

10.1109/IADCC.2009.4808998

Filename

4808998