Assessment of a new acoustic camera localizing vocalizations of newly weaned piglets

Abstract

Objectives: Vocalizations of interest of newly weaned piglets were induced in order to assess the performance of a new acoustic camera, that uses an array of 64 microphones to map sound intensity over video recordings in the form of a heat map. The aim was to determine whether the acoustic camera would be capable of detecting changes in vocalizations of newly weaned piglets, which are related to their affective state. The assessment was performed by (1) determining the types of background noise picked up by the acoustic camera; (2) assessing the sensitivity of vocalization localization for grunts, barks and screams/squeals and (3) determining the minimum distance between two piglets in order for the acoustic camera to localize grunts, screams/squeals and simultaneous vocalizations correctly. Materials and methods: For the first two objectives, six piglets were given free access to a corridor, above which the acoustic camera was mounted in order to record sound. Vocalizations for the third objective were recorded during the isolation of two piglets in the corridor. The occurrence of grunts, barks, screams/squeals and simultaneous vocalizations were measured. Results: The results of the first research aim showed that the acoustic camera picked up a lot of background noise. Of all collected sound events occurring at a piglet’s snout, the proportions of sound attributed to the smacking (0.50) and movement (0.41) of the piglets were especially high. The results of the second research aim showed that the order from highest to lowest proportions of correctly localized vocalizations were as following: barks (0.72), screams/squeals (0.67) and grunts (0.19). The proportion of correctly localized barks was significantly higher than the proportion of correctly localized grunts. And lastly, the results of the third research aim showed that grunts produced by an individual piglet could be localized most accurately at a distance of 40 cm and beyond. Screams/squeals produced by one piglet could be localized most accurately at a distance of 20 cm and beyond. The correct localization of simultaneous vocalizations turned out to improve as the distance between the piglets increased. Due to the small sample size and various other limitations of this study, the results should not be generalized to vocalizations of the entire population of newly weaned piglets. Applicability: Some results indicate that the acoustic camera would be ready to detect changes in a piglet’s affective state by measuring vocalizations. Such as the fact that the camera was fairly good at correctly localizing barks and screams/squeals when there were six piglets in the corridor. The acoustic camera was also able to correctly localize grunts and screams/squeals from a distance of ≥40 and ≥20 cm respectively when there were two piglets in the corridor. Despite these advantages, several improvements are needed before the acoustic camera would be ready for use in commercial farms. First of all, the acoustic camera should be able to filter out more background noise, especially the smacking and movement of the piglets. For a welfare assessment based upon vocalizations, it is also necessary for the acoustic camera to become better at localizing grunts. Future research is necessary, and should focus, among other things, on the vocalization localization of larger groups of newly weaned piglets housed under different conditions, and being able to differentiate between positive and negative vocalizations using the acoustic camera.