BILINGUAL SPEECH RECOGNITION SYSTEM FOR ISOLATED WORDS USING DEEP NEURAL NETWORK

Bilingual speech recognition gradually becomes an attractive research topic because bilingual writings or speech is present in almost everywhere in present scenario. Bilingual speech recognition is the process of recognizing two different languages in the speech signal. In this research it is proposed to develop a bilingual speech recognition for isolated words uttered in English and Tamil language. There are two major modules in the proposed system are, language recognition and speech recognition. Experimental results show that the English utterances of female having higher accuracy (93.6%) than male utterances in language recognition. The performance of system using six hidden layer is higher (approximately 90% for all male and female utterances) when compared to the other hidden layers.