One 5 AM–9 PM window a day — eight full days plus a part-day at each end.
51,282 detections of 125 species across 137 h of audio — 13,624 (27%) at ≥ 0.80 confidence. Every one of the 44 species from deployment 1 turned up again.
A detection is one species in one 3-second window — how the model works.
Detections by hour, pooled across the whole deployment. Effort is uneven — hover any hour for its coverage and per-hour rate.
Detections per day, with the running species total. Hatched bars are the two part-days.
The 41 species that cleared 0.80 confidence at least once — 99.1% of all detections. The other 84 are pooled into the last row: 43 of them are one or two detections at the 0.25 floor, but 7 recur on seven or more days. All 125 are in the data table.
▶ plays that species' best clip, cut from your own recordings · hover a photo (reference) to enlarge · badge = IUCN Red List status.