Survey on Phoneme Recognition using Support Vector Machine

Automatic Speech Recognition (ASR) is a process in which speech signal is converted into a sequence of words, other linguistic units by making use of an algorithm which is implemented as a computer program. The speech recognition system would support many valuable applications that require human interaction with machine. The major objective with which ASR works is the development of the techniques and a system that enables the computers to recognize speech as input. Most precisely speech recognition means phoneme recognition. Good phonetic decoding leads to good word decoding, and the ability to recognize the English phones accurately will undoubtedly provide the basis for an accurate word recognizer. In this work, a detailed survey on a classification technique called support vector machine (SVM) is carried out. Linear SVM are mainly used where linear separation between two classes is possible and in nonlinear SVM's kernel functions are used for conversion purpose. Among two types of SVM classifier soft margin SVM is used to improve the performance when error occurs and least square SVM is used for large scale classification. Citation: Fathima Nazarath P. A. "Survey on Phoneme Recognition using Support Vector Machine." Global Research and Development Journal For Engineering [ERTEE' 19], Adishankara Institute of Engineering and Technology, kalady, Kerala KTU(Kerala Technological University), (March 2019) : pp. 187 - 192.