Haikubox review: High-tech backyard bird

For bird watchers, Being able to identify birds by their song is sacred. Some people seem to have the nature of hearing a song once and remembering it forever. If you’re like me—if you’re not one of those people—you’re wondering, “Why isn’t Shazam available to birds?” You must have thought that. Indeed, if Shazam can identify a song playing on some blown speakers with a few seconds of bad sound, one can figure out how to do the same for a bird song in a nearby tree.

That, in a nutshell, is what the creators of Hyakubox did – Birdsong created Shazam.

That in itself is welcome and surprising, but Hyakubox turns out to be much more than that. It’s one of those rare pieces of technology that enhances your connection to the world around you instead of cutting you off.

Neural network

The bird migration started earlier this year. I know this because Hykubox told me. Not in so many words, but by mid-August, new warblers are starting to announce their arrival, which means they’re already heading south to their winter quarters in Central and South America.

With a full-time job and three kids, I don’t have time to get out and bird every day. I must have missed the battles of Cape May when they passed for two weeks at the end of August. They don’t last long, and I always thought they would get stuck in a birch scrub a good mile down the road. Although they tend to spend their days elsewhere, thanks to Hykubox, I know they’ll be passing by my door in the morning. Hykubox alerts me every time it hears one so I can check them out.

This is the magic of Hykubox – it expands your world.

Photography: Haikubox

Surprisingly, Hyakubox is pretty unassuming on the outside. It is a 4 x 6 inch round-square box about 2 inches thick. At the bottom there is a sealed outlet for the power cord and a small microphone that records sounds around the Hikubox. Although the device is weather resistant and I didn’t experience any problems in the rain, the company advises not to expose it to direct sunlight. Don’t steal it. Once you’ve found a good spot, plug it in and connect it to your Wi-Fi network via the Haikubox Connect app. Hikubox starts recording audio 24/7.

This is the end of the hardware, but that’s not where the magic really lies. Once connected to your wireless network, Hykubox sends its recorded sounds to servers at the world-renowned Cornell Ornithology Laboratory.

The Ornithology Lab has thousands of bird song samples and a neural network to process them. Neural networks are a type of machine-learning software best suited for recognizing patterns in voice — that’s how Siri and Google Assistant understand your voice. Similarly, a neural network can filter out bird songs from background noise. To find patterns, one must first learn what a pattern is. Cornell Library’s Bird Song Recordings Training AI to learn which sounds are birdsong and which are watering the garden.

Cornell has been tinkering with neural networks for some time. If you want to experience this without investing in Haikubox, you can pick up Cornell’s Merlin Bird Identification app, which relies on a smaller dataset and the same AI processor that Haikubox uses. Haikubox creator David Mann told WIRED that Haikubox uses an improved version of BirdNet, called BirdNet for Haikubox.

Neither BirdNet nor BirdNet is perfect for Haikubox, but it is often surprisingly accurate. Better yet, you can use the Haikubox app to help improve the AI.

Multiple views

Haikubox via Scott Gilbertson

You can use the Haikubox app for Android, iOS or the web-based interface to see which birds your Haikubox has heard and attempted to identify. When you first open the app, you set up an account, and then you can sign in to that account on any device. The data in each app is the same, and I used all three during my testing. I found the mobile app more useful for notifications, but since I can open eBird and other additional information in background tabs, I prefer to skip and browse the species information in the web app.

Source link

Related posts

Leave a Comment

seven − 1 =