How? 🤔

Chirps manifest as excursions in the electric organ discharge frequency. To discern the individual chirps in a recording featuring multiple fish separated solely by frequency, we delve into the frequency domain. This involves the computation of spectrograms, ensuring ample temporal resolution for chirp distinction and sufficient frequency resolution for fish differentiation. The outcome is a series of images.

This framework facilitates the application of potent computer vision algorithms, such as a faster-R-CNN, for the detection of objects like chirps within these 'images.' Each chirp detection yields a bounding box, a motif echoed in the package's logo.

Post-processing steps refine the results, assigning chirp times to the fundamental frequencies of each fish captured in the recording.

Still not sold? Check out the demo »