Recognising behaviourally relevant pig vocalisations for welfare assessment via a lightweight deep acoustic model

Key Points

• Pig vocalisations indicate behavioural signals related to welfare, helping assess animal conditions.
• A lightweight deep acoustic model was used to classify vocalisations, achieving high accuracy rates of up to 90%.
• The analysis employed deep learning techniques to interpret vocal patterns, focusing on distinguishing relevant sounds.
• Improving animal welfare assessments can enhance farming practices, indicating a need for broader application of acoustic monitoring.