Skip to main content
. 2021 Aug 3;11:15733. doi: 10.1038/s41598-021-95076-6

Figure 3.

Figure 3

Sound Processing routine. A traditional approach requires four steps from sound to classification (clockwise). The MFCC step may be avoided when using Random Forests19. Future routines based only on CNN may process the sound directly (counterclockwise step).