My colleagues and I have a new paper out in the journal Avian Conservation and Ecology called “Direction-of-arrival estimation of animal vocalizations for monitoring animal behaviour and improving estimates of abundance”. The paper presents the results of a test of a sound localization system made from two Wildlife Acoustics Songmeter SM3 devices, configured so that an algorithm can tell a user the precise direction towards a sound source. Click here to download the paper.
Sound localization is a potentially transformative technology for the field of bioacoustics. Traditional means of data collection involve handheld recording devices pointed at a bird. The primary shortcoming of this method is that it requires a human to be present in the field at all times. For obvious reasons, humans are constrained in the time they can be in the field – they need to eat and sleep, and chances are they need to split their recording efforts between multiple focal birds (my study population for my PhD numbered about 12 birds). The end result is that traditional recording approaches lead us to miss most of what’s happening in a particular area.
Automated recording units (ARU’s), such as the Songmeter SM3, are now becoming a preferred alternative. They can record for hours on end from a given location, and can permanently document the sounds of an area. However, ARU’s come with their own shortcomings. One drawback is that a simple stereo recording gives a listener minimal information about the direction of an incoming sound. Yes, a listener can hear if the bird was singing on the left or right side of the device, but discrimination in the forward-backward or up-down directions is not possible. This makes it difficult to count the number of birds singing, and practically impossible to know anything about what those birds might be doing.
We sought to improve the directional abilities of ARU’s by configuring the SM3 for direction-of-arrival estimation (See picture).
The results were promising: our system was able to identify the angle of an incoming sound with an accuracy within 10 degrees of the true angle, 90% of the time.
We propose that a system such as this one could improve the utility of ARU’s for counting birds and for tracking bird movements. This latter functionality is particularly relevant for my current work, since having some sort of contextual data (in this case, the location of the bird) would add a crucial element to a recording sample and allow for some very interesting analyses of singing behavior in Cassin’s Vireos. This summer, I will be deploying two of these devices on the same bird’s territory, and will use the intersection of the directional bearings to estimate a singing bird’s absolute location, and track their movements around the territory.
My hope is that this technology will help me ramp up my data collection to the next level, and reveal untold insights into the lives of Cassin’s Vireo! Stay tuned for an update as the field season progresses.