期刊名称：International Journal of Computer Science and Information Technologies
电子版ISSN：0975-9646
出版年度：2014
卷号：5
期号：5
页码：6664-6669
出版社：TechScience Publications
摘要：Speech Recognition technology is one of the fast growing engineering technologies. The task of speech recognition is to convert speech into a sequence of words by a computer program. As the most natural communication modality for humans, the ultimate dream of speech recognition is to enable people to communicate more naturally and effectively. The accuracy of automatic speech recognition remains one of the most important research challenges after years of research and development. There are a number of well-known factors that determine the accuracy of a speech recognition system. Several speech recognition algorithms were adapted for different language. In this research continuous speech recognition is concentrated over Tamil language. This paper divides the proposed framework into three stages namely preprocessing, feature extraction and classification. Feature extraction is the process of retaining useful information of the signal while discarding redundant and unwanted information. In feature extraction stage, the proposed framework uses MFCC for transforming the signal into a form appropriate for classification. Classification is done by combination of EWTLBO and HMM. Teaching–Learning-Based Optimization (TLBO) is recently being used as a reliable, accurate and robust optimization technique scheme for global optimization over continuous spaces. This research presents an improved variant of TLBO algorithm, called Enhanced Weighted Teaching–Learning-Based Optimization (EWTLBO). A performance comparison of the proposed method is provided against the original TLBO and some other algorithms. An additional parameter “weight” is introduced to the existing TLBO algorithm to increase convergence rate. To enhance the search space of the algorithm an elitist concept is introduced to improve the performance of existing TLBO algorithm. EWTLBO is used to find the optimization value which is passed through the HMM to get the recognized output.
