Lyrebird
wild birdsong, measured and resynthesized · syrinx model · no samples
Lyrebird measures wild birdsong and sings it back from a physical model of the
syrinx.
It is an art–science instrument by
Shahab Nedaei. Every note is
integrated in your browser from the Laje–Mindlin oscillator, driven by the two
parameters real birds control: air-sac pressure and syringeal tension. No audio is
sampled and none is shipped.
The corpus holds 444,213 measured syllables from
2,489 species, grouped into 15,341 call types and
projected into a three-dimensional embedding that flares where a live sound lands in it.
It does not translate birdsong into language, and it says so in the interface.
Machine learning can characterise a repertoire and link calls to behaviour; it cannot
recover propositional meaning. Every phrase carries a label naming where its claim came
from — playback-verified, documented, inferred, field recording, or invented.
This page needs JavaScript, because there is no server here to do the work: the
measurement, the clustering read-out and the synthesis all run locally. The writing is
readable without it — the paper,
a plain-text summary, and
the source.