ROCCA acoustic classifier for Northeast Atlantic delphinid species

Random Forest models for classifying vocalizations of Atlantic delphinid species, compatible for use within the ROCCA (Real-time Odontocete Call Classification Algorithm) module in PAMGuard. When used together to classify encounters containing multiple vocalizations, this ROCCA classifier was estimated to range in accuracy from 60% for Grampus griseus encounters to 92% for Orcinus orca encounters. Model files can be imported into ROCCA to classify dolphin whistles and clicks from acoustic recordings from the Northeast Atlantic. The following seven species can be identified::

• Atlantic white-sided dolphin (Lagenorhynchus acutus)
• Common bottlenose dolphin (Tursiops truncatus)
• Killer whale (Orcinus orca)
• Long-finned pilot whale (Globicephala melas)
• Risso's dolphin (Grampus griseus)
• Short-beaked common dolphin (Delphinus delphis)
• White-beaked dolphin (Lagenorhynchus albirostris)

Both click and whistle classifiers were trained and evaluated exclusively on visually-confirmed single species recordings of these species and only evaluated on recordings spatiotemporally independent from those used to train the models. ROCCA is currently being used in various government-funded monitoring projects in Scotland and it is our hope that it will continue to facilitate acoustic monitoring of delphinids in Northeast Atlantic waters in the future.

Use ROCCA models easily within the widely used open-source software PAMGuard.

These models are designed to be imported into PAMGuard's ROCCA module to classify individual detections of whistles and clicks made using the Whistle & Moan Detector and Click Detector modules, respectively, to species. These base classifications are then intended to be fed into an eventClassifier, which is currently available as separate graphical-user-interface provided here. Please consult the included README file for specific guidance and tutorials.