This sample data was created using DTrack.
Source audio was generated while playing public domain audio and whistling. Attempts were made to provide a limited set of whistles that roughly resembled previous attempts; outliers were discarded.
Listening to recordings to currate a set of tagged data is trivial but tedious.
This data was collected to snapshot the data collection process, allowing focus on machien learning tasks.
Sourced from Public Domain materials.
The source material for background music should be public domain, although this could not be confirmed, and 2-second clips do not qualify as redistribution. The whistles are arbitrary noises and not copyrightable.
To my knowledge, nothing in this repository can have a copyright or license.