Neural-Edge

Identify the birds around you by their calls. BirdNET runs on the Neural Engine. No internet, no upload, no account. Just listen, name, and log.

NeuralEdge identifies the birds around you by their calls, using a neural network that runs entirely on your iPhone.

Open the app, tap Sentry Mode, and it listens. When a bird sings nearby, the model — Cornell's BirdNET v2.4, accelerated by the Apple Neural Engine — names the species, logs the time and place, and goes back to listening. No internet, no upload, no account.

What NeuralEdge does:

Sentry Mode — continuous on-device listening. Bird calls in, species names out, no cloud round-trip. Bird Log — every detection logged with time, location, and confidence score. Browse by species or by day. Map view — every detection pinned on a map. See which birds frequent which places. Species info — taxonomy from the eBird catalogue and a short summary from Wikipedia. Privacy filter — human speech is suppressed before analysis. Conversations near the microphone are never sent to the model. Webhook — push detection events to a URL you choose (Home Assistant, Zapier, your own server). Off by default.

What NeuralEdge does not do:

No accounts. No cloud. No upload. Detection runs entirely on-device, accelerated by the Apple Neural Engine. The privacy nutrition label is "Data Not Collected" — NeuralEdge has nothing to send anywhere even if it wanted to.

Designed for iPhone (iOS 26). Microphone audio is analysed locally and never stored. Location is GPS-tagged on each detection and kept on your device.

Made by Indy.inc in the United Kingdom.