Human Scream Detection and Analysis for Crime Reduction

Crime is still prevalent in societies across the world, and it increases with each passing day. There are daily reports of different criminal activities that include robbery, murder, rape, aggravated assault, and homicide, but many of them still happen undetected due to delayed or complete police inaction. Such a delay is mostly due to poor address information or failure to notify law enforcement on time. This paper will seek to address these challenges through designing a system that identifies crimes in real-time using Machine Learning algorithms to analyse feeds from social media as well as using audio sensors to recognize distress signals. The system helps in reaching at the crime scene faster by applying the location sharing through the short messaging service alert and so the police stations which are nearest to the crime scene get alert immediately. In field tests, the system helped to decrease reaction time by 40% and rate of crime intervention increased by 30% showing likelihood of boosting public safety.