Abstract:
Climate change and human encroachment are some of the major
threats facing several natural ecosystems around the world.
To ensure the protection of ecosystems under threat, it is important
to monitor the biodiversity within these ecosystems to
determine when conservation efforts are necessary. For this to
be achieved, technologies that allow large areas to be monitored
in a cost effective manner are essential. In this work we investigate
the use of acoustic recordings obtained using a low cost
Raspberry Pi based recorder to monitor the Hartlaub’s Turaco
in central Kenya. This species is endemic to East Africa and
faces habitat loss due to climate change. Using simple features
derived from the spectrograms of the recordings, a Gaussian
mixture model classifier is able to accurately screen large data
sets for presence of the Hartlaub’s Turaco call. In addition, we
present a method based on musical note onset detection to determine
the number of calls within a recording.