Although scientists have yet to decode the ABCs of sperm whales or what an elephant's signals mean, there is considerable activity in studying animal communication. Now it's the orangutan's turn. By combining traditional recording and analysis with artificial intelligence (AI), scientists identified three distinct types of pulses in the primate's long call. According to a study published May 14 in the journal, their voices are likely even more complex. Peer J Life and Environment.
Orangutans are great apes found primarily in Southeast Asia. They are the largest tree-dwelling mammals and feed mainly on insects and flowers. They are also known for complex social behavior and communication. However, as with many species, understanding the nuances of their vocal repertoire has been a challenge.
[Related: Why do humans talk? Tree-dwelling orangutans might hold the answer.]
“Our research aims to unravel the complexities of orangutan long calls, which play an important role in their communication over vast distances in the dense rainforests of Indonesia,” study co-author and Cornell University primatologist Wendy Erb said in a statement. ” “Over the course of three years, we collected hundreds of long call recordings, revealing an interesting array of vocal diversity,” Erb said.
Their long calls are loud and booming sounds produced by males. They are highly variable from individual to individual and are used to communicate over long distances between widely spaced animals. Early studies compiled a glossary of pulse types. Using the same dictionary, Erb and his team tried to determine how many types of pulses they could describe, what characteristics could differentiate them, and see how well they were classified.
Using these traditionally recorded video and audio sources, Erb and team used machine learning to closely analyze the long calls of 13 individual orangutans. He wanted to determine the number of pulse types present in his voice and estimate their classification.
“Through a combination of supervised and unsupervised analytical methods, we identified three distinct pulse types that differed well in both humans and machines,” Erb said. “Although our study represents an important step forward in understanding orangutan communication, there is still much to be discovered. Orangutans may possess a much larger repertoire of vocal types than we have described, which they Highlight the complexity of the acoustic system.
[Related: Artificial intelligence is helping scientists decode animal languages.]
According to the team, this work highlights how complex sounds are in the animal kingdom, noting that pulses aren't just random sounds. It also demonstrates the power of collaborating across scientific disciplines and combining traditional research methods with advances in AI.
“We hope our findings will encourage further exploration of vocal complexity in different species and pave the way for future discoveries in animal communication,” Erb said.