Multimodal system for recording individual-level behaviors in songbird groups

Area of Science:

• Ethology
• Bioacoustics
• Animal Behavior

Background:

• Identifying individuals and behaviors in animal groups is crucial for ecological studies.
• Segmenting rapid behaviors and isolating vocalizations from mixed audio signals presents significant challenges.
• Multimodal recording systems with synchronized data streams offer promising solutions.

Purpose of the Study:

• To design and implement a modular system (BirdPark) for simultaneous multimodal data recording of small animals.
• To overcome challenges in synchronizing data from various sensors for fine-grained behavioral analysis.
• To enable the dissection of complex social behaviors in animal groups.

Main Methods:

• Developed a modular system (BirdPark) utilizing custom low-power frequency-modulated radio transmitters on animals.
• Employed a custom software-defined radio receiver with multi-antenna demodulation to enhance signal quality and eliminate data loss.
• Utilized a single clock for digital acquisition of all data streams, ensuring precise cross-modal synchronization.

Main Results:

• Achieved a 6.5 dB increase in signal-to-noise ratio compared to single-antenna demodulation.
• Enabled reconstruction of wing-stroke phases in free-flying songbirds.
• Successfully separated vocalizations from up to eight interacting birds.

Conclusions:

• The BirdPark system provides accurate cross-modal redundancies for dissecting rapid behaviors below video frame rates.
• This technology facilitates the automated analysis of complex social interactions in animal groups.
• Paves the way for advanced research in animal ethology and bioacoustics.