Recently, a lot of attention has been paid to exploring Artificial Intelligence (AI) for analyzing audio and vocal data, offering a wide range of capabilities in precision livestock farming including poultry behavior monitoring. Animal behaviors provide significant insights into the mental and physical well-being of poultry, serving as an important indicator of their health and subjective states. With the world’s population projected to reach 9.5 billion by 2050 and the demand for animal products like eggs, meat, and milk expected to increase by 70% from 2005 levels, it becomes vital to develop automated, precise systems for monitoring poultry behaviors. This achievement is especially important for managing poultry health and welfare efficiently, overcoming the constraints of manual behavioral observations, which are time-consuming. Automated AI-based systems are thus increasingly becoming crucial for monitoring and promoting good welfare within the growing livestock industry.

In this paper, we aim to develop a simple and efficient AI audio-based approach to recognize chickens’ key behaviors such as eating, greeting, foraging, hunting, and tidbitting to improve poultry farming. First, the proposed study performs cepstral and entropy analyses on the chickens’ vocalizations to extract new vocal features. Second, a simple deep unsupervised clustering method is proposed to recognize the behaviors of the chickens. Alternations in recognized behaviors can be indicators of lameness in chickens. Here, we used an open access chicken language dataset consisting of a total of 74 distinct chicken calls with their probable meanings as based on careful observations. Promising results are obtained by the proposed scheme for chicken behavior monitoring, enabling poultry personnel to accurately determine the health and well-being of chickens.