Identify sound in raucous acoustic environment

Abstract

Due to 2023, over 200 million people worldwide are visually impaired. The needs of people with visual impairments receive scant attention in todays world. Most of them cannot walk independently on Crowded thoroughfares. There are still some challenges in developing assistive devices for the visually impaired. This paper focuses on a classification system within the earphone worn on the ear that can distinguish between different sounds and can be located by the Sharpless of the sound waves. The proposed method comprises two main modules: the first is to transfer the audio signals to Spectrograms, which is done in Python, and then a trained Convolutional Neural Network (CNN) is used in Matlab to identify different types of sounds. When tested in a real-life environment, this system proved useful and accurate in identifying dangerous signals. This innovation is intended to provide them with the optimal time to evacuate dangerous areas, ensuring their safety.