This study investigated context-dependent characteristics of vocalizations in captive bottlenose dolphins (Tursiops truncatus) under different behavioral situations in Korea. The experiment was conducted on July 21, 2025, at the Whale Ecology Experience Center in Ulsan, Korea, involving four captive bottlenose dolphins housed together in a single tank. Four experimental conditions were established: (1) a normal condition without external disturbance, (2) a feeding condition in which a trainer threw food into the tank, (3) a training condition involving responses to trainer signals, and (4) a toy-provided condition. Dolphin vocalizations were recorded for a total of 90 minutes using an Aquafeeler IV A and AQH-200D hydrophone (10 Hz-200 kHz, sensitivity -220 dB re 1 V/μPa) while behavioral data were simultaneously collected using tank monitoring cameras. Acoustic analyses were conducted using PAMGuard software. A total of 197 whistles and 378 buzzes and burst-pulses were recorded during the experiment. In the normal condition, dolphins swam freely and predominantly produced whistles within a frequency range of 6.81-23.82 kHz. During the feeding condition, the production of whistles, buzzes, and burst-pulses decreased while click trains associated with food searching behavior were mainly observed, typically lasting 0.9-1.8 seconds. In the training condition, dolphins primarily produced buzzes and burst-pulses, often accompanied by low-frequency tonal components (1.55 – 8.13 kHz) in response to trainer signals. In the toy-provided condition, dolphins produced clicks associated with object exploration along with buzzes and burst-pulses ranging from 1.3 to 10.8 kHz. Whistles were most frequently recorded in the normal condition accounting for 72.5% of total whistle occurrences whereas buzzes and burst-pulses were most frequent in the toy-provided condition (51.1%). These results demonstrate that bottlenose dolphin vocalizations vary systematically according to behavioral context, suggesting that vocalization frequency and patterns should be useful indicators for inferring behavioral states and have potential applications in passive acoustic monitoring and biological monitoring technologies.