Budapest (Hungary) - The University of Budapest has finished a new study that concluded that it is possible for a computer to analyze and talk back to dogs based on six main kinds of barks.

The software used in the experiment was developed by the Sony Computer Science Laboratory and it looked at 14 Hungarian sheepdogs. Researchers analyzed more than 6,000 barks and were able to classify each one into one of six categories.

The six different barks were called "stranger", a response to seeing a new person in the house, "fight", when the dog's owner encouraged the dog to bark, "walk", in preparation for an outside walk, "alone", when the dog was tied up with no one around, "ball", when the owner was holding a ball or other kind of toy, and "play", a bark used when the owner was playing a game with the dog.

The researchers said the accuracy and ability to distinguish between the different behavioral responses is about on par with their response rates of similar human studies.

"You could use this method to analyze other vocal signals, such as bird songs, or categorize video recordings," said University of Budapest ethologist Eotovos Lorand.

Researchers were able to use this data and in a second experiment correctly identified barks of 52% of a new group of dogs. The group's next project is to analyze the barks of "working dogs", like sheep dogs and hunting dogs.

Source : Tom's Hardware US