'Name that tune' algorithm used to identify signature whistles of dolphins
This is a very cool bit of research which has solved the long-standing problem of how to classify the very complex whistles of dolphins. Dolphin 'language' is so complex that is it very difficult to analyse using automatic detection software (like the kind I use for analysing blue whales songs). Up until now the only really successful way to analyse these very complex whistles was classifying them manually using the human eye, where researchers had to go through hours and hours of data and identify each whistle type. Some acoustic data sets can be thousands or tens of thousands of hours long, so this method was not very time effective at all.
Now, by applying the same kind of software used to recognise music, available on smart phone apps, researchers have been able to automatically identify and classify the signature whistles of dolphins, i.e. the characteristic whistles produced by each individual dolphin.
This is a very exciting step forward in the technology available to bioacousticians!
