Enhancing poultry health management through machine learning-based analysis of vocalization signals dataset

Area of Science:

• Agricultural Science
• Animal Science
• Bioacoustics

Background:

• Global demand for animal protein drives poultry industry growth.
• Early detection of poultry diseases is vital for animal welfare and economic stability.
• Current diagnostic methods are often insufficient, leading to significant losses.

Purpose of the Study:

• To create a comprehensive audio dataset of healthy and unhealthy chickens.
• To facilitate the development of smart technologies for early disease detection in poultry.
• To enable continuous, non-invasive, and automated health monitoring of poultry flocks.

Main Methods:

• Collected audio data from 100 day-old chickens over 65 days.
• Separated chickens into treated and untreated groups for respiratory illnesses.
• Utilized high-fidelity audio recording (24-bit/96 kHz) to capture subtle vocalizations.
• Implemented noise reduction techniques and consistent environmental monitoring.

Main Results:

• The untreated group exhibited audible signs of respiratory illness after 30 days.
• Distinct audio patterns were observed differentiating healthy and unhealthy birds.
• A dataset of chicken vocalizations was successfully compiled in WAV format.

Conclusions:

• Audio analysis of chicken vocalizations can serve as an effective tool for early disease detection.
• The developed dataset supports the creation of automated poultry health monitoring systems.
• This technology has the potential to significantly improve poultry farm management and reduce economic losses.