Crimes like robberies assaults and murders occurring daily crime is a major issue on a global scale. Lack of immediate information frequently makes it difficult for police to react swiftly. This study focuses on using real-time human scream detection to identify crimes earlier. The system will either send a notification or sound an alarm when it detects something serious. With the use of specialized software it can distinguish between normal background noise and a scream. It makes use of machine-learning methods like Support Vector Machines (SVM) and Multilayer Perceptrons (MLP) to produce the most accurate results. They aids in the recognition of more complex sound patterns whereas SVM excels at handling noisy sounds.