Then the observer collects data from the magnitude spectrum and the graph axis showing the frequency. 3 levels of frequency of each audio file will be collected, in which 50 knockings will be performed on each object and the magnitude will recorded according to the frequency applied. For instance, the green steel produces 3 magnitudes from 3 different frequency which are 2480 Hz, 4480 Hz and 5093 Hz. The red steel gives out 2078 Hz, 3757 Hz and 6259 Hz. When the knocking process is done, find the best frequency value to be used for the calculation of the object's sound separation, where we will use the average value produced by each frequency level.