BIRDeep Audio Annotations: Expert-Labeled Bird Song Recordings from AudioMoth Recorders in Doñana National Park, Spain (EBD-CSIC and Universidad de Córdoba)
- Provider
- Estación Biológica de Doñana (CSIC) and Universidad de Córdoba (BIRDeep project)
- License
- MIT (source)
- Size
- 3,587 files
- Format
- WAV, CSV (in sampled files)
- Stored at
- Hugging Face (https://huggingface.co/datasets/GrunCrow/BIRDeep_AudioAnnotations)
- Access
- Free and public. No account with the data provider is needed.
- DOI
- 10.57967/hf/2801
- Category
- Environmental Science
- Source
- https://huggingface.co/datasets/GrunCrow/BIRDeep_AudioAnnotations
About this dataset
Passive acoustic monitoring recordings of bird vocalizations from Doñana National Park in southern Spain, annotated by expert ornithologists of the Estación Biológica de Doñana (CSIC) and the Universidad de Córdoba for the BIRDeep project on deep learning bird song detection and species identification. AudioMoth recorders were placed at nine sites in three habitats (low and high Mediterranean shrubland, ecotone and marshland) from February 2023. About 500 minutes of one-minute recordings, chosen around dawn and morning bird activity, carry 3,749 annotations in 38 classes: individual species, genus-level labels, a general Bird class and a No Audio soundscape class. Each annotation records start and end time, low and high frequency, species, annotator, recorder and date. The repository (about 3.2 GB) has 641 WAV files in Audios/, a matching spectrogram PNG for each clip in images/, YOLOv8 bounding-box label files in labels/ and train, validation and test CSV metadata following Darwin Core terms, so it can be used both as an audio event detection set and as a spectrogram object detection set. Dataset DOI 10.57967/hf/2801. Released by the BIRDeep project under the MIT License; cite the BIRDeep paper, Decoding the Sounds of Doñana: Advancements in Bird Detection and Identification Through Deep Learning.
How to load this dataset in Python
import pandas as pd
df = pd.read_csv("hf://datasets/GrunCrow/BIRDeep_AudioAnnotations/dataset.csv")
df.head()Data structure
File types in sampled files: WAV, CSV
Files
Example file: dataset.csv
Analyze this dataset without downloading it
On Scigantic, BIRDeep Audio Annotations: Expert-Labeled Bird Song Recordings from AudioMoth Recorders in Doñana National Park, Spain (EBD-CSIC and Universidad de Córdoba) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.
This page is also available as Markdown and as JSON.
Related Environmental Science datasets
- Moorings - Estuarine and coastal water quality monitoring - Near real time
- DEA Fuel Moisture Content (Sentinel-2, Australia)
- NEON Tree Crowns, DeepTreeAttention Edition: 41,738 Species-Labelled Crowns at 38 NEON Sites
- GLOBIO 4 Terrestrial Mean Species Abundance (MSA) Biodiversity Intactness, 2015 and SSP 2050 Scenarios
- Multi-perspective plains zebras: four drone-swarm missions on one synchronised timebase
- PUUM passive acoustic recordings: unlabelled Hawaiian bird audio from Pu'u Maka'ala Natural Area Reserve
- OBIS: Ocean Biodiversity Information System (2024)
- Secondary Forest Brazil