Data annotation for spectrogram extraction

[bluemellophone]

We need to be able to extract chirps and pulses from a waveform file, ideally using traditional signal processing techniques to detect and segment each chirp. To do this without any machine learning, we have used qualitative distinctions so far but would rather make data-driven decisions on how to best extract a compressed representation. This effort would require annotating ~30 files for each of the 34 North American bat species (~1,000 files total).

The primary task would be to annotate every pulse and place a start / end point on each detection. Annotators would need to note any harmonics, and echos, or overlapping chirps.