Launching today
Hushh
AI baby cry analyzer that runs entirely on your device
5 followers
AI baby cry analyzer that runs entirely on your device
5 followers
Hushh is a privacy-first AI baby cry analyzer that runs entirely on your device. Using on-device AI, it analyzes baby cries directly in your browser - no cloud uploads, no server-side processing. Install it as a PWA and explore how modern browser AI can help parents while keeping sensitive audio private.
Hushh – An AI-powered baby cry analyzer that runs entirely on your device 👶🔒
Hi everyone! 👋
A few days ago, while waiting to pick up my daughter from summer school, I noticed a mother trying everything she could to calm her crying baby. It reminded me of when my daughter was an infant.
Every parent knows this game:
* Is the baby hungry?
* Tired?
* Needs a diaper change?
* Feeling uncomfortable?
* Or... something else entirely?
Unless babies could talk, we're all just making educated guesses.
That moment made me wonder:
With today's on-device AI models, could we build an app that listens to a baby's cry and suggests the most likely reason?
The privacy requirement immediately became my top priority.
I didn't want parents uploading recordings of their babies to a cloud server. Sensitive data like a baby's voice should stay on the family's own device whenever possible.
So I built Hushh.
## What is Hushh?
Hushh is an experimental Progressive Web App (PWA) that analyzes baby crying sounds using AI entirely on-device.
That means:
✅ No audio leaves your device
✅ No server-side inference
✅ Works directly in your browser
✅ Privacy-first by design
## Why a PWA?
I wanted something parents could open instantly without installing a native app.
Modern browsers are becoming surprisingly powerful, and on-device AI is opening exciting possibilities for privacy-preserving applications.
Hushh is also an experiment to see how far we can push AI inside a browser.
## Tech highlights
* Progressive Web App (PWA)
* On-device AI inference
* Privacy-first architecture
* No cloud processing for baby audio
* Runs locally in the browser
## This is still an experiment
I don't claim that Hushh can replace parental judgment or medical advice.
The goal is to explore whether on-device AI can become a helpful assistant for parents - not a decision maker.
I'd love feedback on:
* Accuracy
* UX
* Performance
* Device compatibility
* Ideas for future features
## Try it
🌐 Live Demo:
https://tryhushh.vercel.app/
💻 GitHub:
https://github.com/nhatnx/hushh
If you're a parent, an AI enthusiast, or someone interested in privacy-preserving machine learning, I'd really appreciate your thoughts.
Thanks for checking it out! ❤️
A pwa that runs offline is already a huge win for tired parents at 3am. One thing that would make it even better is adding a simple timeline view that logs cry patterns throughout the night, so you can spot trends over the week and share that data with your pediatrician without exporting anything manually.
@lamont_shi Thanks! That's an excellent suggestion. I really like the idea of turning Hushh into more than a "single cry analyzer" and helping parents understand patterns over time.
A local, privacy-first timeline that automatically groups cry events and highlights trends (time of day, frequency, possible causes, etc.) would be incredibly valuable—especially during those 3 a.m. moments when everything blurs together. 🙂
I also love your point about sharing insights with a pediatrician. My goal would be to make that as simple as possible while keeping the raw audio on the device by default. A shareable summary or report could be a great addition.
Thanks for the idea—I'm adding it to my roadmap!