ChirosetEurope: A curated bioacoustics dataset of European bat vocalizations

This repository hosts ChirosetEurope v1.0, a curated bioacoustic dataset indexing 11,517 recordings of search-phase echolocation calls, feeding buzzes, and social vocalizations from 35 European bat species across 32 countries in Europe. The dataset was assembled from multiple public repositories and expert-curated collections and encompasses diverse geographic regions, habitats, recording devices, and species groups. Within the constraints of currently available recordings, ChirosetEurope aims to maximize taxonomic coverage, geographic diversity, and annotation quality. It is intended as an adaptable reference resource for comparative bioacoustics, acoustic monitoring, and automated acoustic analysis. Predefined group-aware training, validation, and test partitions are provided for machine-learning applications. To respect the licensing conditions of the original sources, this archive does not redistribute audio files that remain available through public repositories. Instead, the harmonized metadata registry provides persistent source URLs and unique recording identifiers through which users can retrieve those files in accordance with their respective licences and terms of use. Audio files from contributing collections for which redistribution permission was obtained are hosted directly in this archive as audio.zip. The included starter notebook provides an automated workflow for querying and filtering the metadata, retrieving selected recordings from their respective locations, and applying configurable preprocessing operations, including channel conversion, resampling, and segmentation. This allows users to reconstruct and adapt the resource locally according to their research questions and applicable licensing requirements. Dataset documentation The construction, curation, structure, technical validation, and recommended use of ChirosetEurope are described in: Baier et al. (2026), A curated bioacoustics dataset of European bat vocalizations. Example downstream application ChirosetEurope v1.0 was used in a separate study to develop and evaluate an automated classifier for European bat vocalizations: Ghani et al. (2026), ChiroEcho: extending automated bat vocalisation classification beyond the learned taxonomy. Accepted at CV4Ecology 2026.