Process raw audio files using BirdNET to generate species detections, embeddings, and clusters by call type, as well as representative training samples for validation.
Run audio file splitter / renamer Generate BirdNET detections Aggregate BirdNET detections Generate BirdNET embeddings Aggregatae BirdNET embeddings Cluster over BirdNET embeddings by species to call type Sample representative call segments
- 1-minute segments
- 3-second segments
- stratify by call type, confidence quantile (0-50%, 50-75%, 75-87.5%, 87.5%-100%) Export samples to Whombat