IoT based Real-time Voice Analysis and Smart Monitoring System for Disabled People

Abstract

This research emphasizes on Internet of Things (IoT) based affordable platforms to take proper and timely measures for disabled people.It is usually observed that people with different disabilities face difficulties in all walks of life, and adequate caretaking measures are not adopted in most cases.Real time and consistent caretaking for such handicapped people is a tedious task.This paper introduces an IoT based real time analysis and alerting system for the disabled people.The proposed standalone system consistently monitors voice activity of person and in case of any abnormality in analysis outcomes, the system automatically notifies concerned hospital or caregiver to prompt for the patient's situation.The voice features are extracted from analysed voice by employing Discrete Cosine Transform (DCT), and classified through Support Vector Machine (SVM).The prototype has been developed by using Raspberry Pi single board along with voice recording module, Wi-Fi module and LCD Screen.Cloud web services have been used to store the real time activity and performing voice analysis.Montreal Affective Voices (MAV) dataset has been utilized for training and testing of voice recognition.The designed system can be regarded as a rescue system for people suffering from various life threatening health conditions including bipolar disorder, hysteria, cardiac arrest, etc.An accuracy of 81.74% has been achieved for MAV dataset, whereas an accuracy of 67.90% is achieved for real time voice input as depicted in the analysed results.