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!
Peotry error when installing, fixed by removing tensorflow-macos. Maybe an issue in new versions of tensorflow https://github.com/tensorflow/recommenders/issues/713