Inexpensive human audiometric system using Raspberry Pi and artificial intelligence

Abstract

The most common and widespread disease in Iraq is hearing impairment for children and newborns. Also, in cities, people are exposed to high levels of noise, loud sounds at work, like factories, and machinery noise. In this paper, a system was designed and implemented to measure the level of hearing in the human ear, in order to reduce the cost of these devices. This system uses Raspberry Pi 3 microcontrollers, which are considered cheap and have high capabilities in open-source programming. Their abundant availability will lead to the provision of these systems in homes, health centers, and hospitals. In this proposed algorithm, two sine waves are generated by the microcontroller with different frequencies. It is transmitted by the MP3 audio transmission cable through the analog-to-digital (ADC) port. These audio signals are generated at a frequency of (0.5 to 12 kHz), these frequencies are the ones that humans can hear, and they can be represented by pulse width modulator (PWM) technology (x=255 samples). Convolutional neural network (CNN) is trained on the dataset acquired through deep learning algorithms.