Table 1. Parameters and input data used to train each species classifier. Training parameters: parameters input into Kaleidoscope Pro to determine whether or not a sound could potentially refer to the focal species and should be extracted from the ~70,000 hour acoustic dataset. Training data: audio recordings used to train the species classifiers; each training dataset includes vocalizations from the focal species (focal), vocalizations from other similar sounding species found at the same site (non-focal), audio recorded at the study site, and clear recordings from the sound library Xeno-Canto (www.xeno-canto.org). Verified detections: total number of times the focal species was accurately detected in the ~70,000 hour audio dataset, manually verified to remove false positives. The presence or absence of each focal species was manually assessed in 1,000 audio files and compared with classifier outputs to estimate the false negative rate and the recall rate.