Bioacoustics has emerged as a powerful, non-invasive tool for monitoring animal welfare, offering opportunities for automated health and behavioral assessment in swine production. This scoping review, conducted according to PRISMA-ScR guidelines, mapped scientific evidence on using sound analysis to detect indicators of health, pain, stress, and emotion in pigs. Five electronic databases were systematically searched without time restriction, yielding 65 eligible studies. Thematic synthesis revealed that bioacoustics is most frequently applied to pain and welfare evaluation, followed by disease monitoring and behavioral analysis. Key acoustic biomarkers identified include screams for nociceptive pain, grunts for emotional valence, and coughs for respiratory distress. Machine learning and deep learning models demonstrated high predictive accuracy, often exceeding 90%, particularly for automatic cough detection and stress classification. Recent developments in edge computing and TinyML have expanded the feasibility of continuous on-farm monitoring. Bioacoustics represents a practical pathway toward Precision Livestock Farming, enabling early detection of welfare challenges. However, standardization of recording protocols and model validation across diverse production systems remains a priority for successful commercial and field-validated application.
Buchini et al. (Thu,) studied this question.