Pengenalan Suara Terhadap Pergerakan Robot Beroda dengan Metode FFT

Abstract

Speech Recognition to Mobile Robot Movement With FFT Method is a system that serves to move the robot direction based on the given sound. Usually to move a wheeled robot someone is still using the PC or remote control. With these problems researchers create a software system and hardware designed to recognize human voice commands. The voice signal will be processed on PC using FFT (Fast Fourier Transform) method in LabVIEW to find out the amplitude and frequency. Then to know the features of advanced sound, turn right and turn left using MFCC (Mel Frequency Cepstrum Coefficients) method then the result of these characteristics is converted into time domain then stored into reference template by DTW (Dynamic Time Warping) method. The experimental results show that the system can recognize forward sound, turn right, and turn left has an average success rate of 81.1% and has an average error of 18.8%. Sound recognition errors are caused because when the speech sounds forward, turn right, and turn left the environment is not always quiet.