A repo designed to convert audio-based "weak" labels to "strong" intraclip labels. Provides a pipeline to compare automated moment-to-moment labels to human labels. Methods range from DSP based foreground-background separation, cross-correlation based template matching, as well as bird presence sound event detection deep learning models!
add in a helper function that converts a Kaleidoscope-compatible dataframe into a raven compatible folder. Basically, Raven has a text file for each audio clip.
add in a helper function that converts a Kaleidoscope-compatible dataframe into a raven compatible folder. Basically, Raven has a text file for each audio clip.