birdnet_verify_segments: (BETA) Verify BirdNET segment detections

In nationalparkservice/NSNSDAcoustics: Streamline National Park Service Bioacoustics Analysis

(BETA) Verify BirdNET segment detections

Description

(BETA) Interactive function that produces spectrograms and wave clips enabling a user to verify BirdNET "segments".

Usage

birdnet_verify_segments(
  verification.library,
  segments.directory,
  results.directory,
  frq.lim = c(0, 12),
  spec.col = monitoR::gray.3()
)

Arguments

verification.library	Character vector specifying which verification options should be shown to the user. Allows finer control to fit user's needs: user may specify whether a detection is a song, a call, a certain song or call type of interest, false alarm, unsure, or whatever the user needs. This enables maximum flexibility for the user, but also requires some thoughtfulness so that verification options remain consistent. BirdNET provides only species-level confidence and does not classify to song or call types. Thus, depending on the underlying research question and a verifier's expertise and familiarity with a focal species, a user may find themselves in a situation where songs are easily verified, but calls are not. Verification library provides the flexibility to accommodate varying questions and levels of expertise, but must be thought through by the user beforehand.
segments.directory	Directory path for segments.
results.directory	Directory path where you would like the output csv to be placed.
frq.lim	Optional two-element numeric vector specifying frequency limits, in kHz, to apply to the plotted spectrograms. Default = c(0, 12).
spec.col	The colors used to plot verification spectrograms. Default = gray.3(). Spectrogram colors are adjustable, and users may create their own gradients for display. A few spectrogram color options are provided via the R package monitoR, including gray.1(), gray.2(), gray.3(), rainbow.1(), and topo.1(), all of which are based on existing R colors.

Details

This function was developed by the National Park Service Natural Sounds and Night Skies Division to process and explore audio data using BirdNET.

Spectrograms show a segment detected by BirdNET. Title of spectrogram indicates the segment audio file name.

MP3 files create issues in R unless you install 3rd party software (see ?monitoR::readMP3 for details), so this function will operate very slowly on MP3 files since they have to be converted to wave first. For a faster way to deal with verification of MP3 files, consider using the segments.py routine from the command line, as described at the BirdNET-Analyzer Github page.

Value

Returns a CSV of user-input verifications.

See Also

birdnet_analyzer

nationalparkservice/NSNSDAcoustics documentation built on March 4, 2025, 10:24 p.m.