April 16, 2009

Ongoing research on laying hens has showed that bird vocalisations can be reliable indicators of stress forms and stress conditions, according to researchers.

Knowing what and when specific environmental conditions cause stress is the first step to reduce stress in the production environment, said Dr. Michael Darre from the Department of Animal Science, University of Connecticut.

Alleviating stress of poultry flocks is important because it improves bird welfare and production efficiency, said Darre.

Research involving the analysis of chicken vocalisations is non-invasive and provides immediate and accurate feedback on the levels and types of bird stress, he said.

In a wider study called the "Dolittle Project: Classifying Animal Vocalisations", modern speech-processing algorithms for the first time is applied to animal vocalisations.

Darre said one of the primary initial goals of the work was to determine if vocal patterns were the same for all types of stressors and research has shown that they are not.

The analysis was carried out by digitally recording laying hens' vocals under different stressful conditions, and then, after removing background noise, analyses the recordings using a modified version of the Hidden Markov Model (HMM) with voice recognition algorithms.

The modified HMM is a statistical model that represents both the temporal and spectral characteristics of audio signals. When applied to chicken vocals, HMM has achieved an overall condition classification accuracy of 74 percent.

Darre and research partner Ebenezer Out-Nyarko are also looking at how vocal patterns of chickens change from hatching to adulthood.

The goal is to develop a "black box" that can be placed in chicken houses that allows farmers to remotely monitor their birds, notifies the farmers when it detects stress vocalisations, and identifies the type of stress so action can be taken immediately.