Word-Based LID Using HMM and Bi-gram Modeling

Abstract

Language identification is the task of automatically identifying the language of the speech signal uttered by an unknown speaker. An N language identification task is to classify an input speech utterance, spoken by unknown speaker and of unknown text, as belonging to one of the N languages. LID has applications as a front-end for machines of multi-lingual information retrieval system, multi-lingual speech recognition system and speech to speech translation system. In this paper, hidden Markov model is used for speech recognition and language modeling, i.e., bi-gram model which is the special case of N-gram model (n = 2 for bi-gram). The maximum-likelihood classifier is used to identify the language of given test speech.