Washington, January 17 : Hungarian researchers have shown that computer programs can be better than humans at classifying dog barks.

Csaba Molnar, a researcher from Eotvos Lorand University, says that a new piece of software is able to classify dog barks according to different situations.

The software may even identify barks from individual dogs, something that humans find challenging.

He says that his team’s work, published in the journal Animal Cognition, suggests computer programs to be the most accurate tool for studying acoustic communications amongst animals.

Molnar revealed that the software analysed more than 6000 barks from 14 Hungarian sheepdogs (Mudi breed) during the study.

The experiments were conducted in six different situations — stranger, fight, walk, alone, ball, and play.

The barks were recorded with a tape recorder and later digitalized in a computer, wherein they were coded, classified and evaluated.

In the first experiment, looking at classification of barks into different situations, the software correctly classified the barks in 43 per cent of cases. The best recognition rates were achieved for ‘fight’ and ‘stranger’ contexts, and the poorest rate was achieved when categorizing ‘play’ barks.

Based on these findings, the researcher came to the conclusion that different motivational states of dogs in aggressive, friendly or submissive contexts may result in acoustically different barks.

In the second experiment, looking at the recognition of individual dogs, the algorithm correctly classified the barks in 52 per cent of cases. It could reliably discriminate among individual dogs that humans could not.

Based on this finding, the researchers concluded that there are individual differences in barks of dogs, even though humans are unable to recognise them.

Highlighting the value of their new methodology, the authors said: “The use of advanced machine learning algorithms to classify and analyse animal sounds opens new perspectives for the understanding of animal communication… The promising results obtained strongly suggest that advanced machine learning approaches deserve to be considered as a new relevant tool for ethology.” (ANI)