Written and maintained by CASRAI Editorial Board
Last updated
Short answer
Xeno-canto is a global, open-access citizen-science repository and research database dedicated to sharing wildlife audio recordings. Operated by the Xeno-canto Foundation with infrastructure support from Naturalis Biodiversity Center, it provides standardized bioacoustic data, spectrogram visualizations, and Darwin Core metadata for biodiversity monitoring, ethological research, and ecological machine learning pipelines across hundreds of thousands of species records.
Definition and scope
Xeno-canto is an international bioacoustic repository designed to collect, curate, and disseminate recordings of animal vocalizations. Established in May 2005 by Dutch researchers Willem-Pier Vellinga and Robert Planqué, the archive initially documented avian vocalizations in the Neotropics. It expanded into a worldwide repository governed by the Stichting Xeno-canto voor natuurgeluiden (Xeno-canto Foundation) in the Netherlands, supported by technical infrastructure at Naturalis Biodiversity Center in Leiden.
In 2011, Xeno-canto adopted the International Ornithological Congress (IOC) World Bird List as its taxonomic standard to reconcile regional variations. The platform also curates collections for Chiroptera (bats), Anura (frogs and toads), Orthoptera (grasshoppers and crickets), land mammals, and environmental soundscapes. As a primary data publisher to the Global Biodiversity Information Facility (GBIF), its bird collection alone includes over 929,000 occurrence records formatted under Darwin Core standards. Following FAIR data principles, the archive ensures raw bioacoustic data remain findable, accessible, interoperable, and reusable.
How it works or how it is used in practice
Xeno-canto relies on community contributions paired with open peer-review mechanisms. Field researchers and recordists upload wildlife audio files accompanied by required core metadata:
- Taxon: Genus and species per current IOC taxonomy, with optional subspecies.
- Recordist: Full name of the contributor for attribution.
- Geography: Country, locality name, geographic coordinates, and elevation.
- Temporal details: Date and time of the recording.
- Sound type: Classification of vocalization (e.g., song, alarm call, flight call).
Uploads carry a file size limit (originally 1 MB, later increased to 10 MB). The platform automatically generates frequency spectrograms and oscillograms for visual inspection.
Recordings receive quality ratings from A to E. Grade A indicates a high signal-to-noise ratio, close proximity, and minimal background noise. Grades B through E reflect increasing background sound, reverberation, or distant subjects. Lower-quality files are retained because they often represent poorly sampled taxa, rare behaviors, or isolated geographic localities.
Identification dispute resolution operates through open community arbitration. Any user can challenge an identification, temporarily placing the record in a public forum for peer debate. Administrators serve as arbiters who restore or update records once community consensus emerges. Researchers query records via a public REST API or Python libraries like xenocanto-api to filter by species, quality grade, or coordinate bounds.
Types or related concepts
Wildlife sound repositories differ in governance, licensing, and integration with scientific networks. The table below outlines key operational differences:
| Repository | Taxonomic Scope | Licensing Model | Identification Governance | Biodiversity Network Integration |
|---|---|---|---|---|
| Xeno-canto | Birds, bats, anurans, orthopterans, land mammals, soundscapes | Creative Commons (CC BY, CC BY-SA, CC BY-NC-SA, CC BY-NC-ND) | Community challenge and forum arbitration | Direct Darwin Core publishing to GBIF; NLBIF grants |
| Macaulay Library | Birds, amphibians, reptiles, mammals, fishes, insects | Proprietary terms with research access agreements | Institutional verification via eBird editors | Occurrence exports via Cornell Lab pipelines |
| iNaturalist | All organismal taxa (audio, photo, observation) | Customizable Creative Commons or CC0 | Community consensus (two-thirds agreement) | Exports to GBIF for Research Grade records |
| Tierstimmenarchiv | Zoological audio collections | Institutional non-commercial academic licensing | Curator-driven museum taxonomy (MfN Berlin) | Museum biodiversity networks |
These distinctions guide researchers navigating open data licensing frameworks when depositing or synthesizing ecological audio.
Where it matters for labs, clinics or supply decisions
Xeno-canto supports biological research across several core domains:
- Passive Acoustic Monitoring (PAM): Ecological monitoring projects deploy autonomous recording units across habitats. Reference calls from Xeno-canto serve as baseline templates to train automated species detection software.
- Ecological distribution modeling: Georeferenced acoustic occurrences inform species distribution and occupancy models. When analyzing count data across habitat gradients, researchers integrate occurrence metadata into statistical frameworks such as generalized linear mixed models (glmmTMB).
- Machine learning benchmark data: Xeno-canto audio forms the benchmark corpus for annual LifeCLEF and BirdCLEF competitions. Practitioners must differentiate between weak labels (species presence per recording) and strong labels (precise time-frequency bounding boxes), which can be annotated using tools like Cornell’s Raven software.
- Open science data stewardship: Depositing voucher recordings ensures adherence to data management mandates, aligning field protocols with criteria outlined in the RDA FAIR Data Maturity Model and complementing archives like the Qualitative Data Repository.
Common misunderstandings
Common misconceptions about Xeno-canto include:
- Xeno-canto only hosts bird vocalizations: While founded for birds, the repository actively maintains sections for bats, frogs, crickets, grasshoppers, mammals, and soundscapes.
- All recordings are public domain: Audio files are licensed under individual Creative Commons agreements requiring mandatory author attribution; commercial use requires explicit permission for non-commercial licenses.
- Identifications are validated by formal editorial panels: Quality control relies on crowdsourced review; researchers should verify acoustic signatures and review forum threads for disputed records.
- Background species listings represent comprehensive surveys: Recordists identify incidental sounds opportunistically, so background species lists should not be interpreted as systematic community inventories.
Frequently asked questions
What audio specifications does Xeno-canto require?
Xeno-canto accepts audio uploads up to 10 MB per recording, commonly in MP3 format, accompanied by verified locality coordinates, date, and sound categorization.
How do researchers access Xeno-canto data in bulk?
Data can be accessed via the public JSON REST API or downloaded as Darwin Core Archive packages from GBIF containing occurrence metadata and audio links.
Can Xeno-canto audio be used for commercial algorithms?
Commercial re-use depends on the specific Creative Commons license chosen by the contributor. Non-commercial licenses (CC BY-NC) preclude commercial training without direct contributor authorization.
How does the platform handle taxonomic splits?
Following updates to the IOC World Bird List, administrators update taxonomic trees and review locality and elevational data to assign recordings to newly split taxa.
Sources
- Vellinga, W.-P., & Planqué, R. (2015). The Xeno-canto collection and its relation to sound recognition and classification. In Working Notes of CLEF 2015 – Conference and Labs of the Evaluation Forum (CEUR Workshop Proceedings, Vol. 1391). http://ceur-ws.org/Vol-1391/166-CR.pdf (Read 2026-11-08)
- Vellinga, W. P., Planqué, B., & Vellinga, W. (2026). Xeno-canto – Bird sounds from around the world. Occurrence dataset via GBIF.org. https://doi.org/10.15468/qv0ksn (Read 2026-11-08)
- Dufourq, E., et al. (2026). Twelve quick tips for applying deep learning to animal sounds. PLOS Computational Biology, 21(8): e1014604. https://doi.org/10.1371/journal.pcbi.1014604 (Read 2026-11-08)
- Global Biodiversity Information Facility (GBIF). (2026). Xeno-canto Foundation for Nature Sounds publisher profile and occurrence collections. https://www.gbif.org/publisher/1f00d75c-f6fc-4224-a595-975e82d7689c (Read 2026-11-08)
