Dolphins are believed to be one of the most intelligent animal species on the planet -- although precisely how intelligent is difficult to gauge.
That may be about to change. Scientists at the Wild Dolphin Project (WDP) who have been developing a dolphin translator may have succeeded in getting their software to work.
In August 2013, WDP director Denise Herzing was swimming in the Caribbean with a pod of dolphins she has been tracking for 25 years, wearing a prototype of a dolphin translator called Cetacean Hearing and Telemetry (CHAT), developed by the Georgia Institute of Technology's Thad Starner, when one of the dolphin's whistles was translated as the word "sargassum" -- a type of seaweed.
Humans have for some time been communicating with dolphins on a rudimentary level. The animals are capable of responding appropriately to commands and learning to recognise symbols, but how much of that is rote learning and how much the dolphins actually comprehend is difficult to ascertain.
The whistle picked up by CHAT, translated into human speech, was not a whistle from the dolphins' natural repertoire. Dolphins can produce sounds in frequencies inaudible to the human ear, and project sound in various directions without moving their heads, which can make it difficult to figure out which visual cues they are responding to.
Instead, Herzing and her team invented a series of whistles and ascribed them to certain things -- one of which was sargassum -- and trained the dolphins to repeat the whistles when they encountered those things.
Of course, this one instance may not necessarily be significant. It has not been repeated since, and it came in at a higher frequency than the whistle Herzing taught the pod, so there's a possibility the dolphin was just playing with sounds. However, if the dolphins do continue to use the whistles, it could mean that translating their natural language will become much easier, opening insight into dolphin behaviour and communication -- what they are communicating and how.
Although the team's research was cut short last year when the pod moved on, they were able to make some progress. Starner designed algorithms to detect patterns in dolphin whistles that humans might miss, and they were able to identify certain characteristic sounds -- sounds, for example, that were specific to the communication between a mother and her calf.
If the CHAT program succeeds, two-way human-dolphin communication will become a real possibility. It won't work universally, at least to start with -- dolphins "speak" in regional dialects, each of which will need to be translated anew -- but it may be possible to one day know what really goes on in those strange, shiny heads.
Herzing and Starner will present their research in May at a speech and signal processing conference in Florence, Italy.