<![CDATA[In this paper, we focus on speech recognition where speakers used text-dependent which means the text is agreed in advance and will be used next. This system uses MFCC as feature extraction and SVM as pattern recognition. Data were taken from 10 adult speakers with differences in gender, age, and ethnicity. Each speaker provides 50 ballots "computer" and its pronunciation is not controlled resulting in 500 data. Some data training is contaminated with gaussian noise with levels 80dB, 70dB, 60dB, 50dB, 40dB, 30dB, 20dB, 10dB, and 0 dB. The research uses a frame length: of 40 ms, overlapping frames: of 50%, and a coefficient me: of 13. Noise Cancelling was also tested in this research, although not getting optimal results. Pattern recognition SVM with RBF kernel functions produce 100% accurate results. The time process of the Sequential Minimal Optimization algorithm is better than Quadratic Programming algorithms. Increasing the number of speakers to see the performance of the system with a greater amount of data can be made for further research.]]>
Copyrights © 2022
