We have 6+ years of continuous underwater recordings, in which we are currently trying to quantify whistle counts/activity. I have been trying to use Kaleidoscope Pro for whistle detection with signal parameters of 8-15kHz, 0.1-4s and an inter-syllable gap of 0.5s. Results from the detector are then being fed into a trained classifier within Kaleidoscope to identify whistles from among the detections.
My recordings are quite noisy, with sections of broadband noise overlapping with whistles and some whistles are quite faint, both of which are reducing the accuracy in # of whistles detected. Has anyone tried using Kaleidoscope for delphinid whistle detection? Any suggestions for improving detection accuracy?
