Perch · iOS

Who visits
your garden?

Leave your phone by an open window. Perch names the birds it hears and paints them into a collage of your own garden.

205 British birdsOn deviceWorks offlineNo account
Coming to the App Store

What it does

Put your phone on the windowsill and get on with your morning. Perch listens, and every bird that sings gets named and noted — a robin at 6:40, a wren at 7:12. It keeps listening with the screen off, so when you come back you find out who visited while you were making breakfast.

It knows its birds

The listening is done by BirdNET, the bird-recognition model built at the Cornell Lab of Ornithology — the same one researchers leave running in forests. Perch carries a full copy inside the app, so identification happens right on your phone rather than on some server. Which means it works in a field, up a hill, anywhere with no signal — the places birds actually are.

Silence is an answer

Perch would rather say nothing than guess. It knows the birds of Britain, and when it hears something else — a lorry, a doorbell, next door’s conversation — it simply stays quiet. That restraint is the whole game: an app that names a bird for every noise can’t be believed when it names a real one. Every identification carries the score behind it, and when one is plainly wrong you can strike it off — a record you can’t correct is one you stop trusting.

A collage, not a checklist

No tick boxes, no lists to complete. All 205 birds are illustrated in one hand: each generated with Gemini from a photograph of that actual species, so the field marks belong to the bird rather than to an impression of one. The ones you’ve heard nestle into each other’s negative space, sized by how often they visit, so your most faithful regular slowly becomes the biggest bird on the page. The whole app takes its colour from whoever just sang: a goldfinch and a crow don’t look the same. When your collage feels finished, share it as a picture.

Take it for a walk

Start a bird walk and put the phone in your pocket. Perch keeps listening with the screen off and traces where you went, so what comes back is a route with the birds strung along it — how far, how long, and who you passed on the way. It ends when you say so, or by itself once you’ve clearly stopped walking, and the finished walk becomes a card you can send someone.

Where you heard them

Switch location on and Perch remembers where you were for each bird, then draws your mornings on a map — one pin for the bench, one for the towpath, each saying how many species it gave you. And if you actually see the bird, take its picture: Perch keeps your photograph beside the call.

Yours first, then the rest of Britain

The catalogue opens on the birds you’ve actually heard and keeps the other two hundred underneath, ready. Each one has a paragraph about it, a reference recording to play against what you just heard, and a note of how often it turns up in Britain. Smaller collections — garden regulars, the wider catalogue — give you something you can finish. Even the app’s icon is a bird you’ve heard: hear a wren, and you can have one on your home screen.

While you’re not looking

A card on the lock screen shows how the session is going, a home-screen widget shows the last few visitors, and — if you want it — Perch will tap you on the shoulder when a bird lands, photograph included. Either only for a species new to you, or for every bird, depending on whether it’s a windowsill or a walk.

Your garden stays yours

No account, no feed, no leaderboard. What you’ve heard stays on your phone: the history, places, photographs and walks you build in the app. Perch collects limited usage and diagnostic information to help us understand whether it works and improve it. The privacy policy explains how that information is handled.

Support

Questions, problems or feedback about Perch: support@poachapp.com.

Identification by BirdNET, from the K. Lisa Yang Center for Conservation Bioacoustics at the Cornell Lab of Ornithology, used under CC BY-NC-SA 4.0. The 205 bird illustrations were generated with Google’s Gemini image models, each one drawn from that species’ own photograph on Wikimedia Commons, which also supplies the reference recordings; the species text is trimmed from Wikipedia. With thanks to Theodore’s Avian Visitors, where a balcony microphone collaged the birds that passed.