This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass. I built BirdEcho, an AI-powered bird sound identification app that encourages people to step outside and explore nature.
Ever heard a bird singing and wondered what species it was? With BirdEcho, you can record its sound and let AI identify it.
The app also lets you:
The goal is simple: less scrolling, more exploring!
🌐 Live App: https://birdecho.pages.dev Open the app, record a bird singing, or use the built-in sample to try bird identification.
💻 GitHub: https://github.com/shadowe1ite/birdecho The complete source code is publicly available, including setup instructions and model documentation.
I built BirdEcho using React, Vite, BirdNET+, and ONNX Runtime Web.
BirdEcho uses the BirdNET+ V3.0 model to recognize bird sounds. Audio is processed directly in the browser using WebAssembly, without sending recordings to an AI server.
The app uses:
One challenge was running a large AI model inside the browser while keeping the app responsive. I used Web Workers to handle inference and split the model into smaller downloadable chunks.
Openly available AI models made BirdEcho possible without depending on paid AI APIs.
By running AI locally, BirdEcho can identify birds without up their recordings to external inference services.
The application's code is open source under MIT, while the BirdNET+ model has its own CC BY-SA 4.0 license and additional usage restrictions.
I believe AI shouldn't always keep us glued to screens. Sometimes, it should encourage us to explore the world around us.
That's the idea behind BirdEcho. 🌿🐦