Urban sound classification using ANN
K. Solomon Harshavardhan, MAHESH MAHESH · 2022
The importance of the integration of artificial intelligence with the applications in use can be widely seen in today's technology. Audio is a key characteristic in this integration as it plays an important role in critical surveillance systems and driverless cars. The paper aims to provide an urban sound classification model that classifies different sounds appropriately. Most of the existing systems in the market have an efficiency and an architecture that can be points of concern when applied in the market. This paper takes advantage of the deep learning techniques to classify the different sounds in an urban environment. The coefficients extracted using the MFCC technique are used to train the proposed ANN model as the working of the MFCC closely resembles the working of the human ear. The Urban Sound 8k dataset has been used for this paper with 80% of the data allotted for training and the remaining 20% for testing. The achieved accuracy of this paper is 87%. From these results, it is safe to conclude that the proposed approach for sound classification can be efficiently used to develop sound classification and other audio dependent systems which can be implemented in the real world.