BattyCoda 2.0: An open-source annotation and classification platform for animal vocalizations

Animal communication research requires efficient tools for annotating and categorizing vocalizations across diverse species. While existing methods often demand thousands of labeled examples and substantial technical setup, many researchers need streamlined solutions for limited data. We present BattyCoda 2.0, an open-source platform that enables researchers to build custom classifiers for animal vocalizations using small training datasets through an intuitive web interface. The platform supports both supervised and unsupervised classification, allowing researchers to discover natural groupings in acoustic data before defining categories. BattyCoda's modular architecture facilitates the integration of novel classification algorithms and segmentation methods as they become available, enabling researchers to leverage cutting-edge techniques without extensive programming. We demonstrate the platform using bat communication calls, where classifiers achieved robust performance even with limited training data. The system automatically labels high-confidence calls while flagging uncertain classifications for manual review, accelerating annotation. Researchers can customize label categories, upload reference spectrograms, and retrain classifiers as datasets grow. By providing a flexible framework accommodating both hypothesis-driven and exploratory approaches, BattyCoda enables the research community to build acoustic libraries and investigate communication systems in understudied species