Behavior Analysis of Convolutional Neural Network for Environmental Sound

Abstract

Computer recognizing environmental sounds is a challenging and complex problem for a machine and an emerging field of research. In this study, the Convolutional Neural Network (CNN) behavior was analyzed against the environmental sounds. The performance level of the Convolutional Neural Network in identifying the environmental sounds using the parameters that we defined yields an excellent overall accuracy of 96.8%. This gives the model an excellent accurate prediction in identifying the given environmental sounds in the area of machine learning. The lowest accuracy among the group is the door knock, but the accuracy of 95.00%, still considered excellent currently in the field, and thus its parameters are fit for the environmental sounds.