In our lab, we monitor the communication of songbirds that are housed together and recorded with a stationary wall microphone. To know which bird utters which vocalization (source separation problem), we attach miniature backpacks to these birds [1]. Each backpack carries an accelerometer that records vibration signals from the body of the bird. From these recordings, we extract the vocalizations to study vocal learning and elucidate the algorithmic principles governing it.
But besides vocal signals, our accelerometers also record non-vocal low-frequency body vibrations such as the heartbeat and respiration. As biological markers of arousal (heartbeat) and planned vocalization (deep inhalation), these signals might provide valuable insight into the dynamics of vocal behavior and learning.
[1] Ter Maat, A., et al. (2014). Zebra finch mates use their forebrain song system in unlearned call communication. PloS one 9.10: e109334

Project outline:
1. Automatically detect low frequency body vibrations.
2. Use machine learning approaches to classify and characterize different signals.
3. Optional: relate these findings to vocal patterns and/or video-based assessments.

Soonest possible project start: immediately (August 2021)