A system to eavesdrop on marmosets

Abstract

This masters thesis describes developing a custom digital recording system to record the vocalizations and behavior of marmosets, which are small primates native to the northeast of Brazil. Animal behavior scientists have traditionally studied communications between only a 'sender' and a 'receiver'. Animal communications however are hypothesized to occur in communication networks involving more than just a pair of animals. In this project a miniaturized recording system aimed at acquiring data to study the communication networks of the common marmoset is underway. The acoustic recording collar project aims to develop a wearable recording embedded platform for a freely behaving primate. A custom embedded platform utilizing a field programmable gate array (FPGA) has been developed to prototype the system. A hardware description language (HDL) has been used to create the FPGA architecture for the collar application, which in this case is the VHSIC Hardware Description Language (VHDL). Sensors used and developed for this application include a global positioning system (GPS), inertial measurement unit (IMU), and digital MEMS microphone. These sensors provide position and accurate time information, behaviorally related motion information, and the acoustic environment of the marmoset. This data comprises the Behavioral Acoustic Biome of the marmoset. Storage of the behavioral acoustic biome data occurs on a local microSD ash memory card. A printed circuit board of footprint 1.36 by 1.18 inches has been completed and the system will be soon fitted to a 3D printed collar wearable by the marmoset. Demonstration of sensor data logging to the microSD ash has been completed. Other developments of the embedded system are ongoing. Ultimately, fitting multiple wearable devices across a troop of freely behaving marmosets will allow novel studies of communication networks in the common marmoset to be undertaken.