Who is whistling? Localizing and identifying phonating dolphins in captivity

Machine Source Localization ofTursiops truncatusWhistle-like Sounds in a Reverberant Aquatic Environment

“…In the context of whistles in reverberant environments, authors have noted 42 the complications introduced by multipath effects -resulting from the combination of 43 sounds received from both the sound source and acoustically reflective boundaries -to 44 standard signal processing techniques. These complications generally arise from the modest results in relatively irregular, low-reverberation environments where they have 48 been evaluated [29][30][31][32]. In unpublished work, we have achieved similar results.…”

“…In particular, no established method 26 exists for recording the whistles of an entire social group of dolphins so as to reliably 27 attribute the signals to specific dolphins. The general problem of sound attribution, 28 which is encountered in almost every area of communication research, is typically 29 approached in one of two ways: (1) by attaching transducers to all potential sound 30 sources, in which case the source identities of sounds can usually be obtained by 31 discarding all but the highest-amplitude sounds in each source-distinctive recorder, or 32 (2), by using a fixed array (or arrays) of transducers, a physics-based algorithm for 33 identifying the physical origin of each sound, and cameras that monitor the physical 34 locations of all potential sources for matching. 35 While notable progress has been made implementing attached transducers (or tags) 36 to identify the sources of dolphin whistles [25][26][27], shortfalls include the need to 37 manually tag every member of the group under consideration, the tendency of tags to 38 fall off, and the tags' inherent lack of convenient means for visualizing caller behavior.…”

Performance Evaluation of Traditional Signal Processing Methods in LocalizingTursiops truncatusWhistles in a Reverberant Aquatic Environment

“…The first part of this paper is concerned 262 with confirming that these methods, as implemented for our system, perform well for 263 near-ideal impulses; since dolphin echolocation clicks have already been well localized in 264 similar fashions [13][14][15], if at closer sensor-subject distances than we explore here, we 265 would expect decent performance. The second part of this paper is concerned with 266 evaluating these methods on tonal sounds modeled after T. truncatus whistles; they are 267 theoretically expected to encounter difficulties [16] and, when experimentally evaluated 268 individually in less rigorous circumstances, have performed modestly at best, not 269 accomodating studies of acoustic exchanges [18][19][20][21][22][23]. In general, our results were what 270 we expected: while near-ideal impulses were successfully localized, whistle-like tones 271 were not.…”

“…Working to localize bottlenose whistles, other authors have at best achieved modest 57 results in irregular, low-reverberation environments [18][19][20][21][22][23]. While a review of the 58 relationship of all methods to the above framework would be out of the scope of this 59 paper, we note that many of the previous authors and others [16] have noted difficulty 60 localizing dolphin whistles in reverberant environments.…”

“…Ever since mid-20th century studies [1,2] observed that Atlantic common bottlenose attribution, allows for the correlation of a call with gestural and other non-acoustic 23 behavior [5,6], and potentially for the examination of acoustic exchanges among 24 conspecifics [5,8]. In contrast, non-signature whistles are relatively scarce, and by 25 definition are not individually distinctive, which precludes such trivial whistle 26 attribution.…”