Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Assessment of heat stress in turkeys using animal vocalization analysis

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  2018 ASABE Annual International Meeting  1801743.(doi:10.13031/aim.201801743)
Authors:   Longshen Liu, Ji-Qin Ni, Yansen Li, Marisa Erasmus, Rachel Stevenson, Mingxia Shen
Keywords:   Acoustic technology, animal stress identification, animal welfare, sound analysis, turkey vocalizations, SVM

Abstract. Heat stress affects animal health, welfare, and productivity. Animal sound analysis has shown useful as an early warning tool of heat stress in some animal species, but it has not yet been studied in turkeys. The objectives of this study were to develop method for turkey sound analysis and determine whether heat stress can induce specific turkey vocalizations. Forty-eight turkeys were bred in 8 isolated rooms under laboratory conditions for 8 weeks and randomly allotted to heat stress (HS) rooms and control (CON) rooms. Turkeys in the HS rooms were exposed to high room temperature three times during the study. Temperatures in the HS rooms were increased from 21 to 34 °C from 9:00 a.m. to 11:20 a.m., kept at 34 °C for 2 h before lowing back to 21 °C during the tests days. The temperature in the CON rooms was maintained at 21 °C. Turkey vocalizations were continuously recorded. The recorded sounds were analyzed using MATLAB. Fast Fourier Transform (FFT) and Mel-Frequency Cepstral Coefficients (MFCC) were used to process the sound signals in the frequency domain and time-frequency domain, respectively. Support vector machine (SVM) was used as the classifier for sound signal recognition. Zero-crossing rate (ZCR) in the time domain, main frequency (MF) in the frequency domain and 3rd dimension MFCC (3D-MFCC) in the time-frequency domain were found the three best sound features. The accuracy of classification was 88.75% based on SVM and the three selected features. The numbers of turkey vocalizations under HS were 43% more than under CON conditions. Turkeys responded to heat stress within 30 min by producing more vocalizations. The results demonstrated the possibility of using turkey vocal sound monitoring and analysis as an early warning tool for heat stress detection.

(Download PDF)    (Export to EndNotes)