Skip to content

Dataset · DBD-research-group

Large bird sound audio dataset for classification

DBD-research-group/BirdSet

Over 7,200 hours of recordings of nearly 10,000 bird species for training and evaluating audio classification models.

What's inside

This is a large-scale bird bioacoustics benchmark featuring over 6,800 hours of training audio (nearly 10,000 classes) and over 400 hours across eight strictly labeled test sets. Created as a more accessible and larger-scale alternative to AudioSet in its niche, it comes with code to reproduce experiments. The specific file storage format is not specified in the card.

What you can do with it

  • Train multi-label classification models for nature and bird sounds
  • Self-supervised training of audio encoders on large amounts of data
  • Benchmark audio classification models under covariate shift conditions
Tasks
Audio classification
Downloads this month
18K
Likes
33

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("DBD-research-group/BirdSet")