Acoustic Realms · Bioacoustics AI Lab
Acoustic Realms
Deep in the ocean, a humpback's song crosses hundreds of miles; in the misty forest, birdsong is a species' signature.
Here, DSP spectral analysis and convolutional neural networkstrain these sounds into recognizable models — running entirely in your browser.
Chapter I · Deep Ocean
Whale Song Recognition
HUMPBACK WHALE VOICEPRINT
The humpback's song is the most complex acoustic structure in the animal kingdom. Based on real recordings released by NOAA and the NPS, we extract mel-spectral features and train a classifier to tell individual whales apart by their voiceprint.
Chapter II · Misty Forest
Birdsong Classification
BIRDSONG SPECIES CNN
Real field recordings from the public Freesound database (CC0 / CC-BY). Spectrograms are generated and a CNN is trained in the browser; then open your microphone, record a bird, and let the model name it on the spot.