Cough Detection System using TinyML

Abstract

In this research paper, a system for the detection of cough is made with Edge Impulse Studio and Arduino 33 BLE Sense. It is able to distinguish between actual cough and a general undesired signal in the background. Edge Impulse Studio has been utilized in this article for training a large dataset of the various samples for cough as well as undesired noise. A highly efficient & machine learning based TinyML model is designed for determining the resonance of Cough on instantaneous basis. The proposed system has been found to achieve almost 97% accuracy of recognition.

Topics & Concepts