African savannah elephants are known for their rich vocal repertoires, but a new analysis using machine learning has uncovered a subtle social behavior in their deepest sounds. Researchers found that elephants tune their low-frequency rumbling calls to more closely match the calls of other elephants in their vicinity. This suggests the animals are actively imitating one another, not just responding with generic calls.
The study relied on machine-learning algorithms to compare acoustic features of rumbles recorded from different elephants. By analyzing patterns in pitch and tone, the algorithms detected that an elephant's deepest calls shifted to resemble those of a nearby individual. This kind of vocal mimicry is rare in the animal kingdom and has previously been observed in species like dolphins and some birds, but it has been difficult to confirm in elephants without such computational tools.
While the source does not specify how many elephants were studied or the exact context of the calls, the finding points to a possible function: matching a neighbor's rumbles may help reinforce social bonds or coordinate group movement. The authors of the research, as reported by New Scientist, suggest that this imitation could be a form of social alignment. Further work will be needed to determine whether the behavior is deliberate or an automatic response to hearing another elephant's voice.