Quentin Bacquelé

Quentin Bacquelé

@quentinbacquele.bsky.social

PhD Student in Acoustic Communication of Birds 🐦‍⬛, Biogeography 🌍 & Artificial Intelligence EPHE - ENES Bioacoustics Research Lab - CEFE https://quentinbacquele.vercel.app/

242 followers188 following14 posts

Quentin Bacquelé reposted

A new analysis in Science of more than 116,000 bird songs has revealed that the astonishing variety of passerine birdsong is built from just eight basic acoustic motifs, like trills, whistles, and harmonic notes. https://scim.ag/4ft2KBF

scim.agThe global biogeography of passerine songsAlthough bird songs are classic models for understanding the evolution of vocal communication, their global diversity has long made the development of a unifying framework challenging. By analyzing th...
View full thread

Quentin Bacquelé reposted

I had the pleasure to write a commentary in Science about the very comprehensive paper by @quentinbacquele.bsky.social et al. (same issue), on the factors influencing birdsong across the world. Link to my commentary: www.science.org/doi/10.1126/..., originally paper: www.science.org/doi/10.1126/...

science.orgBirdsong diversity across the worldBirdsong motifs are shaped by evolutionary history, biological traits, and environmental constraints
View full thread
View full thread
View full thread

6/ A complex, modulated, information-rich song says more, but it degrades fast as it travels. A simple song says less, but it survives the trip. So every species faces a trade-off between how much a song says and how far it carries.

7/ To test this, we modeled how sound travels through real environments. Dense, humid forest scatters sound and fills the silences between notes with echoes. Simple motifs survive that journey. Complex ones fall apart.

View full thread

5/ Here is the key idea. A song has to do two things at once. Carry information (which species, which bird, how good a singer) and reach a listener who may be far away. These two goals pull against each other.

2/ We analyzed 116,000+ songs from 3,160 songbird species. First we measured the texture of each one: how its pitch slides up and down, and how its rhythm changes through time. We used a tool from auditory neuroscience, the modulation power spectrum.

3/ Then we let a machine learning model group the songs by similarity, with no labels and no target. The songs did not spread out evenly. They piled up around 8 points. We call these 8 poles the acoustic motifs.

You are up to date.