An Approach to Extract Feature Using MFCC for Isolated Word in Speaker Identification System

Abstract

The speech is the prominent and natural form of communication among human being. There are different aspects related to speech like speaker identification, speaker recognition, Automatic speech recognition(ASR), speech synthesis etc. The purpose of this work is to study speaker identification system using Hidden markov Model (HMM).The goal of Speaker Identification System (SIS) is to determine which speaker is speaking based on spoken information. The system uses Mel Frequency Cepstral Coefficients(MFCC) for feature extraction , HMM for pattern training and viterbi techniques. The success of MFCC combined with their robust and cost effective combination turned them into a standard choice in speaker identification system.HMM and viterbi decoding provide a highly reliable way of recognizing odia speech.