Review of the Methods for Automated Recognition of Dogs Emotional State Using Acoustic Signals

This paper addresses the problem of creating a tool for categorizing dogs' vocal expressions into emotional states. Animal vocalizations can now be automatically classified using artificial intelligence algorithms. With the help of an examination of the correlation between the regular and irregular components of the signals, the researchers have discovered various acoustic patterns. In this study, we examined the effectiveness of training methods for bark classifications in compliance with the emotional state of the dogs. People evaluated dogs' emotional states. Furthermore, we discovered a significant link between the values predicted by the assessors and the calculated valence and arousal levels. Notable characteristics of dog vocalizations were defined together with analyzing possible specific and individual features of dogs’ acoustic signals. These features were formed from the digital recording of the barks. Different communicative situations were taken accordingly. This research provides useful insights for creating more accurate tools for the classification of emotions from acoustic data and could be applied to enhancing human-machine interfaces.