Automatic Speech Recognition for Indonesian using Linear Predictive Coding (LPC) and Hidden Markov Model (HMM)

Abstract

Speech recognition is influential signal processing in communication technology. Speech recognition has allowed software to recognize the spoken word. Automatic speech recognition could be a solution to recognize the spoken word. This application was developed using Linear Predictive Coding (LPC) for feature extraction of speech signal and Hidden Markov Model (HMM) for generating the model of each the spoken word. The data of speech used for training and testing was produced by 10 speaker (5 men and 5 women) whose each speakers spoke 10 words and each of words spoken for 10 times. This research is tested using 10-fold cross validation for each pair LPC order and HMM states. System performance is measured based on the average accuracy testing from men and women speakers. According to the test results that the amount of HMM states affect the accuracy of system and the best accuracy is 94, 20% using LPC order =13 and HMM state=16.