Sound Source Localisation of Calling Birds in Outdoor Environments
Bioacoustics concerns the analysis of sound produced by animals. It is an important part of conservation research as it is non-intrusive, yet has the potential to inform wildlife managers about the species present in an area. Acoustic recordings units (ARUs) are often used to record the soundscape of an area. Following post-processing, these recordings can be analysed statistically to estimate call rate abundance. However, the lack of information about location of the calling bird means that this cannot be turned into population abundance. Time difference of arrival at multiple synchronised microphones enables the direction of arriving sound to be computed. This can assist in localising calling animals. However, unlike in indoor environments, propagation of sound in the bush can be hard to predict given the effects of wind, terrain, direction the animal was facing in, etc. In this thesis I investigate whether a practical application of sound source localisation of calling birds is possible with current technology. In particular, I investigate (a) the automatic identification of calls from the species of interest, (b) computation of sound direction from a small microphone array and low-power computational system, and (c) sound propagation in the New Zealand forest. As a case study of using the completed system we located a North Island Brown Kiwi incubation burrow based on the calls of the male as he leaves the burrow in the evening. The results show that the method of locating kiwi burrows using the automatic call identification and direction calculation of the system has potential, but there is further work needed to make it practical.