Identification of Four Major Dialects of Assamese Language Using GMM with UBM

Abstract

The Assamese language is spoken by the people of Assam, which is located in India’s north-east corner. The Indo-European language family includes the Assamese language. The pronunciation, grammar, and vocabulary of Assamese are vary in different sections of the state, resulting in different regional dialects of the language. There are four major regional dialects of the Assamese language, namely Central Assamese spoken in and around Nagaon district, Eastern Assamese dialect spoken in the Sibsagar and its neighboring districts, Kamrupi dialect spoken in Kamrup, Nalbari, Barpeta, Kokarajhar and some parts of Bongaigaon district and Goaplari dialect spoken in the Goaplara, Dhuburi and part of Bongaigaon district. Therefore, to develop a universal Assamese speech recognition system that seamlessly recognizes the words spoken in the Assamese language and its dialects, the identification of the dialect is a necessary condition. Using the Gaussian Mixture Model (GMM) and the Gaussian Mixture Model with Universal Background Model, this research proposes a novel technique for recognizing Assamese dialects (GMM-UBM). To extract spectral information from collected voice sample, the Mel-Frequency Cepstral Coefficient (MFCC) is used. Modeling is done using the GMM and GMM-UBM modeling techniques.