Overview

• Forschungsinstitut Nutztierbiologie (FBN) launched a four-year KI-Tierwohl project in 2025 with partners including Fraunhofer IGD and regional universities and institutes.
• The system converts barn audio into spectrogram images so image-focused AI methods can classify vocalizations.
• The dataset spans more than 7,400 individual calls from over 400 pigs across 19 defined situations.
• A farm use case envisions ceiling microphones providing 24/7 analysis, issuing alerts when animals fare poorly and tracking longer-term welfare.
• Researchers cite the barn ‘cocktail‑party effect’—noise and overlapping calls—as a key hurdle, so broad field deployment remains a later goal.