These models are based on the samples provided as free in a DVD from a Computer Music Magazine's issue (The metadata of the samples indicate that they are from 2006)
https://www.musicradar.com/news/tech/free-female-vocal-samplesThe special thing about these is the fact that has three dynamics with the same phrase/melody (soft, medium and hard), so you will get three vocal models with the specific dynamic of it, and you can choose between them.
Using harvest for pitch extraction and trained with 150 epochs each model.
The source of these samples (like which library come from) or even if the singer was credited in the first place is completely unknown.
