Related Experiment Video
Updated: Sep 16, 2025
09:32
Evaluation of Auditory Brainstem Response in Chicken Hatchlings
Published on: April 1, 2022
3.2K
AI-Powered Vocalization Analysis in Poultry: Systematic Review of Health, Behavior, and Welfare Monitoring
Venkatraman Manikandan1, Suresh Neethirajan1,2
1Faculty of Computer Science, Dalhousie University, Halifax, NS B3H 4R2, Canada.
Sensors (Basel, Switzerland)
|July 12, 2025
Summary
Artificial intelligence and bioacoustics are revolutionizing poultry welfare monitoring using vocalization analysis. This review highlights deep learning advancements and edge computing for real-time insights, despite challenges in data and interpretability.
Area of Science:
- * Integrates artificial intelligence (AI) and bioacoustics for non-invasive poultry welfare assessment.
- * Focuses on advanced vocalization analysis for precision livestock farming.
Background:
- * Traditional acoustic feature extraction methods are evolving.
- * Deep learning models offer enhanced capabilities for analyzing complex poultry sounds.
Purpose of the Study:
- * To systematically review the evolution of AI in poultry bioacoustics.
- * To identify current applications, challenges, and future research directions.
Main Methods:
- * Systematic review of literature on AI and bioacoustics in poultry.
- * Bibliometric co-occurrence mapping and thematic clustering analysis.
- * Examination of traditional and deep learning approaches (CNNs, LSTMs, wav2vec2, Whisper).
Main Results:
- * Deep learning models show significant potential for emotion recognition, disease detection, and behavioral phenotyping.
- * Edge computing via TinyML addresses scalability in commercial settings.
- * Methodological bottlenecks include data standardization, inconsistent evaluation, and limited interpretability.
Conclusions:
- * AI-powered bioacoustics is crucial for ethical, transparent precision livestock farming.
- * Explainable AI (XAI) is vital for trust and regulatory compliance.
- * Future work requires multi-modal integration and domain-adaptive models for broader applicability.
