Skip to contents

A sample BirdNET selection table generated from audio recordings collected from 5 sites, in John Prince Research Forest, British Columbia, Canada, during May–June 2023. Recordings were scheduled continuously over 24 hours using a duty cycle of 1 minute on, 4 minutes off. Audio files were analyzed using the BirdNET GUI (model version 2.4) with a confidence threshold of 0.1; all other parameters were kept at their default values.

Usage

example_jprf_2023

Format

example_jprf_2023

A data frame with 392,300 rows and 14 columns. Key columns include:

filepath

Full path to the audio file from which the detection was made

start

Start time of the detection in seconds

end

End time of the detection in seconds

scientific_name

Scientific name of the detected species or sound event

common_name

Common name of the detected species or sound event

confidence

BirdNET-assigned confidence score (0 to 1)

lat

Latitude of the recording location (set to -1 if not embedded)

lon

Longitude of the recording location (set to -1 if not embedded)

week

Week number of the year (set to -1 if not available)

overlap

Whether overlapping segments were analyzed (0 = no, 1 = yes)

sensitivity

Sensitivity parameter used during detection (default = 1)

min_conf

Minimum confidence threshold used during detection (e.g., 0.1)

species_list

Species list constraint used (e.g., "None" if not set)

model

Name of the BirdNET model used for detection

Details

This dataset is useful for demonstrating typical BirdNET output and testing workflows involving species filtering, thresholding, and temporal subsetting.