Abstract:
A fundamental challenge for current research on speech science and technology is under standing and modeling individual variation in spoken language. Individuals have their
own speaking styles, depending on many factors, such as their dialect and accent as well
as their socioeconomic background. These individual differences typically introduce
modeling difficulties for large-scale speaker-independent systems designed to process
input from any variant of a given language. This research focuses on automatically
identifying the dialect or accent of Afan Oromo language speakers given a sample of
their speech, and demonstrates how such a technology can be employed to improve Afan
Oromo Speech Recognition.
In this thesis, we describe a variety of approaches which are used to develop Automatic
speech recognitions and dialect identifications. In particular, we examine stochastic
(statistical) approach or Hidden Markov Model (HMM) which is used to develop Afan
Oromo Dialect and Accent Identification in this study.
In this study, a self-prepared corpus is used and the developed dialect and accent
identifier is built on medium data size collected from specified target dialect area. The
language model used is a bigram language model.
Finally, the experimental result of this study is evaluated using word error rates (WER)
and the performance achieved for the developed model is 41.45% Word Error Rate.
