<![CDATA[The guitar is a musical instrument that has a chord as a reference tone. It is a fact that is not all human auditory system can distinguish between high and low tones of a musical instrument in good accurate. Then, in this research we develop a voice guitar to cord conversion using resilient backproagation neural network (RBNN) as to classifier and Mel Frequency Cepstrum Coefficient (MFCC) as feature extraction. We record 345 for each cord (totally we have 8640 recording data with WAV format). Experiments are conducted for some number of cepstral coefficients (13, 26, and 39), with 100 millisecond as time frame and 40% overlapping between successive frame. Total number of hidden neurons in RBNN model in this experiments are 10, 25, 50 and 100. According to the experiment, the maximum accuracy is 96.88% for 52 number of cepstral coefficients and 100 neurons hidden.]]>
Copyrights © 2013
