Over 180 ornithological societies and organizations around the world are dedicated to the study of birds, and among many things, how human encrochment on their environment is affecting them. Two of the most prominant of these in North America are the National Audubon Society, with over 450 chapters local to just the United States, and the USGS Bird Banding Laboratory, who have banded over 77 million birds since 1920. But why do we care so much about birds? Birds are considered a keystone species, making if possible for many ecosystems to survive and thrive. Birds also help humans immensely, helping keep pests at bay in our agricuture, cleaning animal carcases which in turn reduces rabies transmissions from ferral animals, and much, much more.

To help scientists that study birds, Dr. Paul Flikkema is working with a research group at Politechnico di Melano, Italy, to develop an artificial intelligence tool that identifies birds based on audio recordings of bird vocalizations. However, Dr. Flikkema needs a diverse library of real-world audio recordings with environmental noise to train this tool. Current tools for collecting wildlife audio recordings are either expensive and complicated, or are cheap and lack useful functionality. The most popular of these tools is the AudioMoth because it is relatively cheap and easy to use. The AudioMoth has flaws, however. While it is open source, the hardware and firmware is not well-documented, it's not designed to be modular, and maintenance is required nearly daily to swap SD cards and change batteries, making it demanding to manage a large collection of sensors.

Dr. Flikkema came to us to do research into the possibility of using Silicon Lab's Thunderboard EFM32GG12 to create a new tool that is capable of recording bird vocalization audio and bringing it into MATLAB (a statistical analysis tool) for the user to analyze. You can see the initial description for this project here.