Automatic speech recognition by cuckoo search optimization based artificial neural network classifier

Abstract

People prefer to interface with one another utilizing speech. Since this is the most common method of communication, people also need to interface with machines utilizing speech. Based on this, automatic speech recognition has gained up a big momentum in recent years. In this paper, an efficient automatic speech recognition (ASR) system using Cuckoo Search Optimization (CSO) based optimization technique for Artificial Neural Network (ANN) is presented. Here CSO is used in order to improve the classification performance of neural network. The proposed technique consists of three stages preprocessing, feature extraction and classification. Preprocessing is done in order to remove the background noise present in the speech signal. Then from preprocessed signal two kinds of acoustic features are extracted they are Mel Frequency Cepstrum Coefficients (MFCC) and Linear Predictive Coding Coefficients (LPCC). Finally, these features are given to neural network for training and based on this features the text correspond to the given speech signal is recognized. The proposed method is implemented on MATLAB working platform. The experimental results indicate that ANN used along with CSO algorithm provides best results compared with ordinary ANN. From the results, we can see that our proposed method is well suited for the speech recognition system.