Machine learning helps to discover sperm whale ‘alphabet’

Researchers have cracked the code to a sperm whale ‘alphabet’ after using machine learning technologies to understand vocalizations.

In a published paper titled ‘Contextual and Combinatorial Structure in Sperm Whale Vocalisations,’ the sequences of clicks produced by the sperm whales were evaluated by researchers from MIT Computer Science & Artificial Intelligence Laboratory (CSAIL) and Project CETI.

These clicks all serve different linguistic functions and are known as coda. Each with its combinations of rhythm, tempo, rubato, and ornamentation features.

The findings show how these sound elements are freely combined, helping whales to communicate a huge repertoire of distinguishable codas. The function of each sound isn’t yet known, but the results “show that the sperm whale communication system is, in principle, capable of representing a large space of possible meanings…”

The team says this is similar to those used by “human sound production and representation systems like speed, Morse code, and musical notation.”

The research into the sperm whale language was made possible through the use of machine learning which was deployed to analyze a dataset of 8,719 codas from the Eastern Caribbean region.

Like in how human speech differs, with people from other regions having different accents and intonation, the sperm whale ‘phonetic alphabet’ was found to be similar.

“[The sperm whale alphabet] shows how a small set of axes of variation (place of articulation, manner of articulation, and voicedness in humans; rhythm, tempo, ornamentation, and rubato in sperm whales) give rise to the diverse set of observed phenomes (in humans) or codas (in sperm whales).”

In a conversation with TechCrunch, MIT CSAIL director Daniela Rus explained how machine learning was utilized: “We would get the inputs, and then we adjust our machine learning, to visualize better and to understand more.

“And then we would analyze the output with a biologist.”

“We believe that it’s possible that this is the first instance outside of human language where a communication provides an example of the linguistic concept of duality of patterning. This refers to a set of individually meaningless elements that can be combined to form larger meaningful units, sort of like combining syllables into words.”

Featured Image: Photo by Ray Harrington on Unsplash

The post Machine learning helps to discover sperm whale ‘alphabet’ appeared first on ReadWrite.