prakulkhebbur/omnisense ? reverse-engineered prompt

Build me a working emergency triage app called OmniSense that listens to incoming calls, pulls out the important details in real time, and ranks the most urgent cases first.

I want a simple operator dashboard that shows active calls, their severity, location, and status, plus alerts when a bunch of similar calls start coming in from the same area. It should be able to handle both text and audio input, turn speech into text, and then use that to score each call so a cardiac arrest or active fire jumps to the top ahead of minor issues.

Please make it feel like a live emergency control room, with updates pushing to the screen automatically. Include a way to simulate incoming calls so I can test it without real audio. If you need current docs for the speech to text or realtime parts, look them up online and use the best approach.