Human Scream Detection

Abstract

Crime, including incidents like homicides, assaults, and robberies, is a common problem worldwide and a big concern for society. A common issue is that police often arrive at crime scenes too late, partly because they don’t get timely, accurate information. To help address this, a disguised desktop application is proposed. This program uses modern technologies, including machine learning and deep learning techniques like Support Vector Machines (SVM) and Multilayer Perceptron (MLP), to quickly recognize and analyze human sounds while quietly running in the background. In an emergency, the program automatically sends SMS alerts to chosen contacts. This advanced technology improves accuracy in detecting threats and speeds up response times by identifying specific human sounds amid background noise. This project introduces a way to help control crime by detecting and analyzing human screams in real-time. Using machine learning and deep learning, the system can accurately pick out distress sounds even with background noise and quickly notify authorities. It works by processing audio, identifying key features, and using algorithms to tell apart different types and intensities of screams. When it detects a scream that signals danger, it sends alerts to nearby law enforcement, including the location and audio recording for evidence. The goal is to make communities safer and reduce the negative effects of crime by strengthening people’s confidence in their ability to protect themselves and their neighborhoods.