Ian Agranat from Wildlife Acoustics told me last week, “Yeah… They all start by building their own recording stations… Then they come to see me.” And indeed, his company builds a quite popular wildlife sound recorder. It’s compact, light and has up to 18 days of autonomy. Frontier Labs in Australia also sells a great module.
But it’s hard to help it… Engineers got to engineer… And so we started to look for cheap equipment to record the wild. On our first tries we borrowed equipment from acoustics consultants Modyva. They use Svantek sound level meters with external batteries that let the meters measure the sound pressure level for two days straight. However, it becomes a bit more awkward to record WAV continuously and as these systems are meant for anthropogenic noise, the sampling frequency is limited. It was still good enough for a few days of tests in the Vosges pine forest (pictures below). We positioned an equipment case by a tree with a huge woodpecker hole in it. A black woodpecker visited it, but did not drum (admittedly, it was summer, which is not the drumming season). The recording picked up its contact call though.
Although the Svantek is typical acoustics engineering equipment, a lot of data is captured with it in natural areas, for example during measurement campaigns near wind turbines. That data contains rich wildlife information .
A few months later, we bought a Zoom H1. Connected to a couple of 9 volt batteries, it would last for a couple of days using light recording options (mid-quality mp3). We tested it in a tropical environment (Mamoni reserve, Panama). It was again not the season for woodpeckers (rain, rain, rain…) and the recordings came back empty, but the Zoom survived the humidity quite well. To this day we are still able to use it.
From that, we moved on to bigger things. We bought a RØde omnidirectional microphone, and most importantly, a Raspberry Pi 2 (the 3 was not out yet). The Raspberry is a small Linux computer; it has USB ports where it is easy to plug any sort of equipment, like a sound card, a SIM card dongle… etc. Our recording station architecture is sketched here:
Every morning the Sleepy Pi (Arduino-based power board) starts the Raspberry Pi, which launches the sound recording. Data is saved to a USB stick. Using GNU Octave, we can do calculations on the data (like downsampling) before saving it. Twice a day, the SIM dongle sends a text message to let us know the station is still operating. We powered the whole system with car batteries, which was a rather poor choice as they don’t like full charge/discharge cycles and died on us pretty quickly. Before that unfortunate end, the autonomy was around 10 days. For the future we need to find something lighter, that lasts longer and of course is appropriate for the Belgian weather.
Here are pictures of us on the roof of the university testing the equipment this January. Don’t be fooled by the sun, it was cold and windy! The laptop is used to access the Raspberry Pi (connection through ethernet cable + Putty).
A box for the batteries, a box for the electronics (weatherproof):
The microphone body is inside the electronics box, hopefully kept warm by the heat of other equipment. We protected it with a latex membrane (made from the gloves of our cleaning lady…) and some foam as a windscreen. (Yes, we tested its response to a white noise source). We had humidity-absorbing gel in the box, and I just learned you can make the packages good as new by putting them in a microwave.
Here is the station in the field:
Great results this time around, including more than 2000 woodpecker drums. Unpublished to date but presented at the fabulous Ecoacoustics 2016 conference at Michigan State University .
 Florentin J, Fauville B, Gérard M, Moiny F, Rasmont P, Kouroussis G and Verlinden O. 2015. Soundscape analysis and wildlife presence in the vicinity of a wind turbine. Proceedings of Euronoise 2015; Maastricht, The Netherlands.
 Florentin J, Kouroussis G, Verlinden O. 2016. An Autonomous Recording Station that Searches for Woodpecker Drums. Ecoacoustics 2016, East Lansing, USA (MI). Abstract.