AURA: Automated Understanding & Response Assistant

Currently, the Lebanese Civil Defense process to receive emergency calls is almost entirely manual. When an emergency call is made to the Civil Defense, a dispatcher takes the call and, using pen and paper, takes note of all the important information concerning the emergency, like its nature and location. The dispatcher then uses a localization application, primarily Google Maps, to search for the area, locate the closest Civil Defense station to it, then inform the concerned team through WhatsApp messaging. Finally, for record-keeping, the same dispatcher fills out a digital form with the same information. In order to provide statistics and analytics, a team of 10-20 people manually check the filled forms and perform different kinds of analyses on them. This process is slow, error-prone, and cognitively demanding, especially during high-stress fire emergencies where delays directly affect response effectiveness. This project proposes a real-time transcription and assistance tool designed to improve the dispatching process and in turn provide faster and more reliable decision making. The system incorporates live audio streaming, voice activity detection, and a speech-recognition model to automatically transcribe incoming calls and organize essential information such as location and incident description. All extracted data are stored in a database, enabling dispatchers to act more efficiently while reducing reliance on handwritten notes and manual data transfer. This tool improves information accuracy, reduces stress on emergency dispatchers, and therefore enhances the overall responsiveness of fire emergency operations.

Project Details

[photo]