WoW-Bench shows current audio-language models are far worse than humans at fine-grained acoustic perception tasks on out-of-distribution marine mammal vocalizations.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
WoW-Bench: Evaluating Fine-Grained Acoustic Perception in Audio-Language Models via Marine Mammal Vocalizations
WoW-Bench shows current audio-language models are far worse than humans at fine-grained acoustic perception tasks on out-of-distribution marine mammal vocalizations.