Inspiration
New parents and busy NICU nurses lose sleep and face extreme stress trying to guess what a crying baby needs. We wanted to create a tool to instantly decode baby cries and help support everyone in the household or hospital ward who has to listen to those cries. To stop the late-night guesswork and protect everyone's peace of mind.
What it does
Decodes Cries: cryOS takes acoustic baby audio feeds and calculates exactly what an infant needs with a 98% confidence rate. Predicts Sleep: It maps out tiredness indexes to find the perfect nap windows. Actionable Help: Instead of just sending a loud alert, it gives real care suggestions like swaddle checks. Privacy-First: All processing is done locally, meaning zero audio data ever leaves the room.
How we built it
The Frontend: Dark-mode user dashboard using React.js, Vite, and TypeScript. The UI Assets: Designed an Acoustic Signature Simulator with vector baby faces that warp their expressions in real-time to show comfort levels. The Data Pipeline: Built a system where a user uploads a .wav file, triggering our API endpoint to instantly process the data and update global metrics.
Challenges we ran into
Handling Noise Floors: Traditional monitors get overwhelmed by background noise like loud nursery white noise machines or busy hospital equipment. Dynamic Calibration: It was challenging to build a built-in telemetry tool that accurately measures the ambient noise floor to establish a clean baseline for sound processing. Real-Time Data Flow: Making sure the uploaded audio pipeline passed data fast enough to update the React frontend state smoothly without lag.
Accomplishments that we're proud of
High Accuracy: Hitting a 98% confidence rate on matching acoustic signatures to actual baby needs during testing. Seamless UI Transitions: Getting the vector facial expressions on the dashboard to change states dynamically based on incoming sound files. True Local Privacy: Building a system that functions safely on a local network without relying on shaky cloud uploads.
What we learned
NICU Market Fit: We learned that cryOS is the perfect size and function to fit compact incubator hubs where automated distress tracking is completely missing. Component Lifecycle Management: Deepened our skills in handling live data streams and state rendering inside complex React components.
What's next for cryOS
Smart Voice Filters: Upgrading our system to include true vocal isolation filters that actively strip away loud surrounding hospital and machine noise. Hardware Integration: Moving our code directly onto small physical edge devices like a Raspberry Pi camera and microphone kit. Long-Term Analytics: Adding predictive tracking features to model long-term health, digestive, and sleep cycles over time.
Log in or sign up for Devpost to join the conversation.