International Journal of Speech and Audiology
2025, Vol. 6, Issue 2, Part A
Acoustic classification of canine Vocalizations using spectrogram-based machine learning to infer emotional states
Author(s): Mahi Shah, Shriya Nadagouda and Aanya Naini
Abstract: Dogs use barking as their primary form of communication, conveying emotions and needs much like humans use speech. However, misinterpretation of these vocalizations often results in harmful corrective measures, such as shock collars or devocalization surgeries. (Ehasni et al. 2018) This study developed a machine learning-based system that analyzes dog barks through spectrogram data and classifies them into emotional subtypes, including fear, aggression, and playfulness. A model trained using Apple’s Create ML achieved an overall accuracy of 93.2%, with class-specific accuracies of 93% for fear, 95% for aggression, and 97% for playfulness. These results suggest that machine learning can provide a reliable and humane alternative for interpreting canine emotions, potentially reducing the use of harsh behavioral interventions.
DOI: 10.22271/27103846.2025.v6.i2a.78
Pages: 01-06 | Views: 755 | Downloads: 545
Download Full Article: Click Here
How to cite this article:
Mahi Shah, Shriya Nadagouda and Aanya Naini. Acoustic classification of canine Vocalizations using spectrogram-based machine learning to infer emotional states. International Journal of Speech and Audiology. 2025; 6(2): 01-06. DOI: 10.22271/27103846.2025.v6.i2a.78
Other Journals