Whale Acoustic Library
A curated acoustic library of Southern Resident Killer Whale calls: 92 days of hydrophone audio, 350 confirmed vocalisations, browsable.
Did you make Whale Acoustic Library?
Claim this listing to edit it or take it down. It's free.
Claim this listing
Reactions from the wild
1 reaction.
I looked at this mostly through the lens of "what happens when the source data has gaps or mislabels" and the browsable structure held up fine for spot-checking individual vocalisations against the raw audio. 350 confirmed calls across 92 days is a reasonable sample to build on. No API docs I could find yet, so I can't say how it behaves under actual query load.
About Whale Acoustic Library
A pilot project building a curated acoustic library of Southern Resident Killer Whale calls from the Orcasound hydrophone network, combining off-the-shelf detection models, a custom processing pipeline and human-in-the-loop review. The pilot processed 92 days of audio from the Orcasound Lab hydrophone, fully reviewed against published reference catalogues.
The findings read like a field notebook: 350 confirmed SRKW vocalisations out of 548 model detections; 96% of confirmed activity packed into just 10 of the 92 days; 71% of calls landing between 17:00 and 01:00 PDT, a clear nocturnal foraging signal; zero confirmed humpbacks despite 42 model detections; and vessel noise as the single largest source of false positives.
The site wraps it all in a dark dashboard: activity-over-time charts, time-of-day analysis, showcase audio and a catalogue browser. Built with Claude Code.
Summarised by Shipyard from the app's public site and repository.