ECOSoundSet: a finely annotated dataset for the automated acoustic identification of Orthoptera and Cicadidae in North, Central and temperate Western Europe

2025-04-29

AI Summary

Existing acoustic identification tools for European insects suffer from limited data scale and insufficient ecological representativeness, necessitating a high-quality, broad-coverage, finely annotated, and cross-regional acoustic dataset. To address this, we introduce EuroOrthoCicada—the first high-precision orthopteran and cicada acoustic dataset tailored to temperate Europe—comprising 10,653 field recordings spanning 223 taxonomic units (200 Orthoptera species and 24 Cicadidae species, including subspecies) across Northern, Central, and Western Europe. We propose a novel dual-labeling paradigm: weak labels (species presence) and strong labels (precise time-frequency localization of calls), jointly standardized by an international panel of taxonomic experts. The dataset is stratified into train/validation/test splits (8:1:1) for immediate model development. Empirical evaluation demonstrates that EuroOrthoCicada substantially improves deep learning models’ robustness to faint and overlapping insect vocalizations in complex natural soundscapes and enhances their cross-regional generalization capability.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Search and Optimization: Evolutionary Computation
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Security and Privacy: Data transparency and provenance

Abstract

Currently available tools for the automated acoustic recognition of European insects in natural soundscapes are limited in scope. Large and ecologically heterogeneous acoustic datasets are currently needed for these algorithms to cross-contextually recognize the subtle and complex acoustic signatures produced by each species, thus making the availability of such datasets a key requisite for their development. Here we present ECOSoundSet (European Cicadidae and Orthoptera Sound dataSet), a dataset containing 10,653 recordings of 200 orthopteran and 24 cicada species (217 and 26 respective taxa when including subspecies) present in North, Central, and temperate Western Europe (Andorra, Belgium, Denmark, mainland France and Corsica, Germany, Ireland, Luxembourg, Monaco, Netherlands, United Kingdom, Switzerland), collected partly through targeted fieldwork in South France and Catalonia and partly through contributions from various European entomologists. The dataset is composed of a combination of coarsely labeled recordings, for which we can only infer the presence, at some point, of their target species (weak labeling), and finely annotated recordings, for which we know the specific time and frequency range of each insect sound present in the recording (strong labeling). We also provide a train/validation/test split of the strongly labeled recordings, with respective approximate proportions of 0.8, 0.1 and 0.1, in order to facilitate their incorporation in the training and evaluation of deep learning algorithms. This dataset could serve as a meaningful complement to recordings already available online for the training of deep learning algorithms for the acoustic classification of orthopterans and cicadas in North, Central, and temperate Western Europe.

Problem

• Limited automated acoustic recognition tools for European insects
• Need for large, ecologically diverse datasets for species identification
• Lack of finely annotated insect sound recordings for deep learning

Innovation

• Finely annotated dataset for insect acoustic identification
• Combination of weak and strong labeled recordings
• Train/validation/test split for deep learning algorithms