Low-resource Sinhala Speech Recognition using Deep Learning

Abstract

Automatic speech recognition has progressed considerably in the past several decades for most of the European languages, but still it is a prominent research area for most of the low-resourced languages. This project presents a study to build an automatic speech recognition (ASR) system using the Kaldi toolkit for the Sinhala language which is one of the low-resourced languages with a large lexical variety. We experiment on different deep neural architectures like pre-trained DNN, DNN, TDNN, TDNN+LSTM to enhance the acoustic modeling process and each of their performances is investigated in this document. A statistical model of GMM-HMM is also trained on the same data set keeping it as the baseline model for comparing the effectiveness of deep learning approach. For the language model, a corpus containing more than 20K sentences taken from UCSC LTRL is used to generate the 220K extended lexicon. The experiments are conducted using a phonetically balanced training data set consisting of twenty-five hours of speech data collected from fifty females and twenty males and tested on 1.6 hours of speech data. We present an overview of different architectures with their procedures and compare and contrast the performances of models with the statistical baseline approach. The results obtained show that Deep neural network architecture exceeds the statistical baseline model with a Word Error Rate (WER) of 7.48% on the test data set. The best observed lowest WERs are produced by the TDNN architectures.