Phonic bestiary
There are many things that go bump in the night, and more that put up an awful clamor. It would be interesting to put a microphone outside, pipe it to a computer, and have the computer identify what’s out there. Many animals make distinctive, repetitive calls of the sort that should be recognizable by fairly simple audio analysis. The cricket, cicada, and frogs, to name a few.
So put a microphone outside, pipe the data in by cord, or wireless internet, and record and analyze it. Run it first through Fourier analysis and then identify duration/frequency patterns. Calls with a certain pitch lasting for a certain interval. Some calls are more complex, a chord, a pattern of notes, etc. Calls range from simple insect calls to complex bird calls. Then run repetitive patterns found through a species identification algorithm.
The output would indicate what animals were detected and at what times, and perhaps record audio samples of each type. This sort of data collected routinely each night would give a dynamic picture of the fauna of an area. Crickets at night, birds at daybreak.
If many people used this software and sent data to a central collection point, it could be used to monitor animal range and abundance, and perhaps indicate aspects of regional and year to year variation by detecting what species appear and when each emerges after winter.
Some work has already been done on the software end of this. Here are papers on bird calls, bird calls, and insect recognition.