Automated acoustic recognition for dummies: combination of two user-friendly machine learning approaches optimize species detection: Data repository
Description
1. Dataset contains a .txt file ("Software settings.txt") with the settings introduced in both software employed (Kaleidoscope Pro and BirdNET), and two separate folders: "Validation dataset" and "Field dataset". The "Validation dataset" folder contains all audio recordings (folder "Acoustic recordings") analyzed using both software and manually (.wav format). The raw database ("Raw_database.csv") showing whether the presence of the American toad was detected with each of the three evaluated methods (Kaleidoscope Pro, BirdNET, human scanning) is included. A .txt file ("Metadata.txt") explaining the different variables included in the Raw database is also included. The "Field dataset" contains the raw database ("Raw_database.csv") obtained after applying the two-steps approach described in Methods. It includes whether the presence of the American toad was detected by Kaleidoscope Pro, whether it was detected by BirdNET, or by any of the two software combined. A .txt file ("Metadata.txt") explaining the different variables included in the database is also included.
Files
Steps to reproduce
The acoustic recordings (located within the "Validation dataset" folder) were scanned with both Kaleidoscope Pro (v5.4.7) and BirdNET (v2.2). Settings introduced in Kaleidoscope Pro and BirdNET are detailed in "Software settings.txt". The "Raw_data.csv" files of each folder provide the raw databases generated after scanning both acoustic datasets, the validation dataset and the field dataset.