An open-source bio-logger for studying cetacean behavior and communication.

Area of Science:

• Marine bioacoustics
• Robotics
• Machine learning

Background:

• Bioacoustics research on marine life is expanding.
• Animal-worn devices offer richer data than fixed recorders for understanding marine mammal communication.
• Multimodal datasets are crucial for deep learning applications in marine biology.

Purpose of the Study:

• To present an open-source, non-invasive bio-logger for collecting synchronized audio and sensor data from marine animals.
• To facilitate the scaling of multimodal dataset collection for deep learning.
• To enable rapid prototyping of new recording capabilities.

Main Methods:

• Developed an open-source bio-logger with four suction cups, three synchronized hydrophones, GPS, and sensors for pressure, motion, orientation, temperature, and light.
• Hardware and software are open-source, with designs and code available online.
• Validated performance through lab tests (adhesion, pressure, battery life, audio sensitivity) and field deployments on sperm whales.

Main Results:

• The bio-logger withstood pressures equivalent to 560m, with a battery life of 16.8 hours.
• Achieved audio sensitivity of -205 dB re FS/μPa with a 96 dB dynamic range.
• Field experiments recorded sperm whale vocalizations and behaviors across 10 deployments, totaling 44 hours of data and dives up to 967m.

Conclusions:

• The open-source bio-logger platform advances the study of marine animal biology and communication.
• It integrates robotics, bioacoustics, and machine learning for enhanced data collection.
• This tool supports deep learning applications and rapid prototyping in marine research.