ScreamAlert: A TinyML-Powered Wearable for Instant Acoustic Emergency Detection

ScreamAlert is a TinyML-powered wearable system designed to detect acoustic emergency signals such as human screams in real time. Existing safety systems suffer from delayed response times due to manual reporting and insufficient acoustic intelligence. This paper proposes an integrated solution combining a sound sensor, a machine learning model trained in Python on acoustic features, and dual NodeMCU (ESP8266) microcontrollers communicating via the ESP-NOW protocol. The primary controller captures audio signals and applies ML-based classification to distinguish emergency screams from background noise. Detected alerts are wirelessly transmitted to a secondary controller which presents real-time status on a 16×2 LCD display. Experimental evaluation demonstrates accurate scream detection with ultra-low communication latency, without requiring internet connectivity or cloud infrastructure. The system contributes toward affordable, accessible, and deployable wearable emergency detection technology.