Paper Key : IRJ************420
Author: Prof. Usha Kumari V,K Karthik,Chaithra K,Bharath S,K Chetan
Date Published: 01 Apr 2025
Abstract
Ensuring personal safety in unpredictable situations is a growing concern, especially when individuals cannot manually activate an SOS alert. This project introduces an Audio-Based SOS System that autonomously detects distress signals, such as screams, using deep learning and real-time audio processing. A convolutional neural network (CNN) extracts features from ambient audio and classifies distress signals. Upon detection, an automated SOS alert is sent to emergency contacts via WhatsApp, SMS, or email, including real-time location data and a timestamp. The system leverages spectrogram analysis and CNN-based architectures for accurate classification and operates seamlessly on mobile devices using Flutter and Python. By integrating AI-driven distress detection, mobile communication, and real-time tracking, this solution enhances emergency response times, providing a proactive and reliable safety mechanism.