Pengenalan Kata Berbahasa Indonesia dengan Hidden Markov Model (HMM) menggunakan AIgoritme Baum-Welch

Abstract

Speech recognition is the process of converting an acoustic signaJ, captured by a microphone or a telephone, to a set of II·O/·ds.Speech can be defined as waws of air pressure created by airflow pressed out of the lungs and gOing out through the mouth and nasal cavities. The air passes through the vocal fo/ds (chords) via the path from the lungs through the vocal tract, vibrating them at different frequencies. To make 0 computer system reacts as a human being in recognizing a word IS not an easvtask. A good model is needed to represent the speech signal as the input of the speech system. . . This research used Baum-We/ch training algorithm 10 train IlJf.H as the model of a word The purpose of th IS research IS to implement HAl,\! using Baum- We/ch training algorithm to recogni:e an isolated word. lI'ords of this research are ranged tnto 2 types of svllable: theyare :l svllables and 3 syllables. Speaker of this research is also ranged lilia 2 tratned woman speaker and l Irained men speaker. theretore thrs system is said to be speaker-dependeni. In general this research resulted same 11.\1,\1,.that rcprescnt specch signal input as all Indonesian word. The best H\f,\f la recogni:e all isolated word IS I/MM IISlIIg 3 hidden states that lI'ere trained IIP 10 JOepochs and the best accuracy IS 8J J:l5%.