Inspiration

My dad is Deaf, and I noticed that surprising him can be easy. When I call him from another room or approach him suddenly, he sometimes jumps or drops his phone. This made me realize how much of hearing we take for granted: things such as hearing people's footsteps before they're actually in front of us, hearing names being called.

That's why I thought of Sentinel: a way to help my dad stay aware of his surroundings calmly. Instead of getting startled, he can receive a clear visual alert and decide how to respond.

What it does

Sentinel helps people stay alert to their environment by listening for important sounds and creating helpful notifications. The current prototype detects three levels of activity:

  • General noise
  • Attention phrases
  • Urgent events, like alarms

It labels events by urgency and sends notifications through Discord. The system processes audio locally and does not save recordings. It helps users notice what is happening and understand its importance.

How i built it

I built Sentinel with Python, a Raspberry Pi, and a microphone. Python manages audio capture, event detection, and notifications. I created shell scripts to simplify Raspberry Pi setup.

The system processes live audio input, identifies key sound patterns, and turns detections into organized alerts. It uses thresholds and cooldown periods to prevent excessive notifications. Discord was chosen for notifications because it is simple and reliable.

Challenges i faced

The main challenge was deciding what deserves an alert. A microphone can pick up many sounds, and too many alerts would be annoying. I had to balance sensitivity and usefulness while minimizing false alerts.

Sound classification was difficult due to noisy environments. I tested different models and thresholds to find what works best instead of relying on the first solution.

Accomplishments we’re proud of

I am proud that Sentinel became a working prototype rather than just an idea. It started with a personal issue in my family, and I overcame technical challenges to create something useful. The prototype distinguishes between ordinary sounds, attention attempts, and urgent alerts, which has been very rewarding.

What I learned

This project showed me that testing is as important as building. It’s easy to find a solution that works once, but real-world testing can reveal weaknesses.

I learned to compare approaches and make informed decisions about sound classification. A useful product must work in real-life situations, not just in perfect testing environments.

What’s next for Sentinel

The main goal of Sentinel is to help my dad receive important information without being startled.

Looking ahead, I envision Sentinel as an environmental-intelligence platform. It could connect microphones, cameras, and alarms to create a clearer picture of events. This would help individuals know when someone is trying to get their attention and help businesses link signals for improved responses.

Sentinel aims to close the information gap, helping people understand what is happening before they react.

Share this project:

Updates

Submission history