Acoustic localization of crows in pre-roost aggregations

Crows are highly intelligent and social creatures. Each night during the non-breeding period, they gather on large pre-roost aggregations as they move towards their communal roost where they sleep. Crows make numerous and varied vocalizations on these pre-roost aggregations, but the purpose of these calls, and vocal communication in general, in these pre-roost aggregations is not fully understood. In this paper, an array of four microphones is used as a non-intrusive means to observe crow vocal behavior in pre-roost aggregations in the absence of human observers. By passively localizing animal vocalizations, the location of individuals can be monitored while simultaneously recording the acoustic structure and organization of their calls. Simulations and experiment are undertaken to study the performance of two time difference of arrival-based methods (hyperbolic location estimator and maximum likelihood estimator) for call localization. The effect of signal-to-noise ratio and uncertainty in measurement on the localization error is presented. By describing, modeling, and testing these techniques in this innovative context, the authors hope that researchers will employ the authors' approaches in future empirical studies to more fully understand crow vocal behavior.