Date of Award
Doctor of Philosophy (PhD)
Dolphin communication research is an active period of growth. Many researchers expect to find significant communicative capacity in dolphins given their known sociality and large and complex brains. Moreover, given dolphins’ known acoustic sensitivity, serving their well-studied echolocation ability, some researchers have speculated that dolphin communication is mediated in large part by a sophisticated “vocal” language. However, evidence supporting this belief is scarce. Among most dolphin species, a particular tonal class of call, termed the whistle, has been identified as socially important. In particular, for the common bottlenose dolphin, Tursiops truncatus – arguably the focal species of most dolphin cognitive and communication research – research has fixated on “signature whistles,” individuallydistinctive whistles that seem to convey an individual’s identity to conspecifics, can be mimicked, and can be modulated under certain circumstances in ways that may or may not be communicative. Apart from signature whistles, most studies of dolphin calls concern group-based repertoires of whistles and other, pulse-form call types. However, studies of individual repertoires of non-signature whistles, and the phenomenon of combined signature and non-signature vocal exchanges among dolphins, are conspicuously rare in the literature, tending to be limited by either extreme subject confinement or sparse attributions of vocalizer identity. Nevertheless, such studies constitute a logical prerequisite to an understanding of the communicative potential of whistles. This absence can be explained by a methodological limitation in the way in which dolphin sounds are recorded. In particular, no established method exists for recording the whistles of an entire social group of dolphins so as to reliably attribute them to their vocalizers. This thesis proposes a dolphinarium-based system for achieving audio recording with whistle attribution, as well as visual behavioral tracking. Towards achieving the proposed system, I present foundational work involving the installation of permanent hydrophone arrays and cameras in a dolphinarium that enforces strict animal safety regulations. Attributing tonal sounds via the process of sound localization – estimation of a sound’s point of origin based on the physical properties of its propagation – in a highly reverberant environment is a notoriously difficult problem, resistant to many conventional signal processing techniques. This thesis will provide evidence of this difficulty, and also a demonstration of a highly e↵ective machine-learning-based solution to the problem. This thesis also provides miscellaneous hardware and the pieces of a computational pipeline towards completion of the full proposed, automated system. Once completed, the proposed system will provide an enormous data stream that will lend itself to large-scale studies of individual repertoires of non-signature whistles and combined signature and non-signature vocal exchanges among an invariant group of socializing dolphins, representing a unique and necessary achievement in dolphin communication research.
Woodward, Sean Fisher, "Who Said That? Towards a Machine-Prediction-Based Approach to Tursiops Truncatus Whistle Localization and Attribution in a Reverberant Dolphinarium" (2019). Student Theses and Dissertations. 529.