A 60 DOLLARS RASPBERRY PI NAMES EVERY BIRD IN YOUR GARDEN BY SOUND — 6,000 SPECIES
A Raspberry Pi 4, a 15 dollars USB microphone and Cornell Lab's neural network turn your garden into a catalogued aviary — no cloud account, no subscription, just a web dashboard that logs every species by song.
by Nachtzuster
ScienceAI
- difficulty
- ●●○○○
- time
- an evening
- license
- CC BY-NC-SA 4.0
- repo
- repo ACTIVE1,059 stars
●●○○○ · an evening · CC BY-NC-SA 4.0 · 1,059 stars · repo ACTIVE
WHAT YOU’LL NEED
- Electronic parts — full list with part numbers in the repo BOM
- Dev board / microcontroller — runs the project firmware
Partner
COMPAREE VERDICT
BirdNET-Pi is exactly what the title says: Cornell's bird song classifier running entirely on a Raspberry Pi with a cheap USB mic plugged in. The install script is one line, the web dashboard shows spectrograms and species confidence scores in realtime, and the whole thing runs 24/7 without any cloud dependency. The hardest part is not the software — it is microphone placement. Put it under an eave and you will log every car and lawnmower; put it in the wrong spot and you will miss the birds two metres away. The second gotcha is the non-commercial licence, which means this stays a personal project. If you want to learn what actually visits your garden, or contribute detections to citizen science networks, this is a weekend well spent. If you expect plug-and-play results without touching sensitivity settings or reviewing false positives in the first few days, set that expectation lower. The single thing most likely to go wrong: underwhelming results because the microphone is in the wrong place or the Pi is too far from the router for stable recording.
IN THE REPO
NOT IN THE REPO
- —Complete installation script and docs in the repository — works on Raspberry Pi 3B+, 4 and 5.
- —No printed parts, no soldering — this is a software project with a USB microphone and optional case.
- —BirdNET model trained by Cornell Lab is included; supports 6,000+ bird species worldwide.
- —Community fork by Nachtzuster; original project by Patrick McGuire still exists but this one sees active updates.
- —Licence is CC BY-NC-SA 4.0 — you cannot sell it or use it commercially.
- —Reddit and forum support is strong; the install script handles model downloads and the web server automatically.
Can I build this?
Build at your own risk. Projects involve tools, electronics and sometimes mains voltage — follow the creator’s safety notes.
Partner · KickstarterCORE POWER BANK S: 0.2" thin, 2.8 oz, 5000 mAh with 22.5W fast charging. Semi-solid battery, carbon fiber, IPX7 waterproof.
Videos
Community walkthrough — not official project video.
Gallery
Start here
Navigation into the creator’s own docs — we don’t rewrite the guide, we route you to the source.
- 1.Flash Raspberry Pi OS Lite to a microSD card (32GB minimum) and enable SSH before first boot.(The install script works on full Raspberry Pi OS too, but Lite is faster and uses less power.)
- 2.Run the one-line installer from the README — it pulls the BirdNET model, sets up the web server and starts listening. (The script takes 20–40 minutes and will reboot the Pi automatically at the end.)
- 3.Open the web dashboard at the Pi's IP address and review the first hour of detections.(Expect false positives in the first session — planes, wind, dogs. You will tune confidence thresholds in the settings to filter them.)
- 4.Move the microphone based on what you see in the logs — closer to feeders, away from roads, out of direct wind.(This is the step that makes or breaks the project. The neural network is excellent; microphone placement is on you.)
Resources
Documentation, files and community threads for this build — we link straight to the original sources and never rehost the creator’s files.
- Reddit discussion: detections in Raleigh, NC reddit.com
- Build video youtube.com
- Project site birdnetpi.com
KNOWN ISSUES
- The USB microphone matters more than you think — a £10 lapel mic works, but mounting it inside a plastic case will muffle everything. Test it outside the case first.
- Non-commercial licence means you cannot sell dashboards, run this as a paid service or integrate it into a commercial monitoring product.
- The Pi needs stable power and reliable network — if it reboots during a detection, that log is gone. Use Ethernet if you can.
- Confidence thresholds ship conservative; you will get false positives (leaf blowers labelled as woodpeckers) until you tune them per species.
- The BirdNET model is trained globally but weighted toward North America and Europe — rare or regional species may be missed or misidentified.
- If you mount this outdoors, weather sealing is your problem — the Pi itself is not rated for moisture, and condensation will kill an SD card.
Does this work at night?
Yes — the model includes nocturnal species and owls, and the Pi runs 24/7. Detections during the day are far more common unless you live near an active owl population.
Can I use this to identify calls I recorded on my phone?
Yes, with manual effort — the system is built for live monitoring, but you can upload audio files to the server and process them through the same model. The interface is not designed for batch uploads.
What is the difference between this fork and the original BirdNET-Pi?
Nachtzuster's fork sees active updates and community support; the original by Patrick McGuire still exists but is less frequently maintained. Functionally they are nearly identical.
How much bandwidth does this use?
Almost none — all processing is local. The only network traffic is you accessing the dashboard or the Pi fetching model updates (rare, and only if you enable them).
Can I export the data?
Yes — the dashboard includes CSV export, and detections are stored in an SQLite database you can query directly if you want raw access.
Community builds
No community builds yet — be the first, we feature the best ones.
Discussion1
FROM THE COMPAREE TEAM
The model knows 6,000+ species but the quality of your log depends entirely on where you put the microphone — under an eave, next to a feeder, away from the road. Where would you mount yours, and what would you expect to catch first?
Nachtzuster
Nachtzuster maintains the most active community fork of BirdNET-Pi, building on Patrick McGuire's original implementation of Cornell Lab's BirdNET neural network for Raspberry Pi. The project is open-source and supported by a community of birders and makers who contribute fixes, feature requests and regional model improvements.
Keep building — related projects
Same category, picked by how close they are to this build.
SOFTWARE ONLYAn AI inbox triage that drafts replies but physically cannot hit send
AIAutomation●●○○○starter
SOFTWARE ONLYAn AI agent that reads every invoice and never pays one
AIAutomation●●○○○starter
SOFTWARE ONLYAn AI agent that drafts a reply to every review and physically cannot post one
AIAutomation●●○○○starter
SOFTWARE ONLYAn AI support agent that answers only from your docs and physically cannot make one up
AIAutomation●●○○○starter
DISCLAIMER
- Comparee is not the author of the projects featured here. All rights to each project belong to its creator — every page links to the original source, and we never host creators’ files.
- Information is provided without warranty and may become outdated as projects evolve. Prices are indicative bands only — always check the creator’s parts list for current costs.
- Building and operating any project is at your own responsibility. Protective equipment, safe workshop practice and compliance with local regulations are the builder’s responsibility.
CompareeTEAM24d agoedited
Practical notes from our verification: the Nachtzuster fork is the one with active GitHub issues and recent commits; the original McGuire repo still exists but this is the branch the community uses. There is no single official video — the walkthroughs are all community-made, and quality varies. The biggest practical surprise was how much microphone placement matters: the neural network is excellent, but a mic stuck inside a plastic case or pointed at a road will log traffic instead of birdsong. Budget an afternoon for tuning after the one-line install, not just the install itself.