Voice Model and Dataset Attribution

This text-to-speech system uses several open-source voice synthesis models and datasets:

VITS (Variational Inference Text-to-Speech) model architecture, developed by Thorsten (based on Kim et al., 2021), which is open source and permissively licensed.

The en/vctk/vits model trained on the VCTK Corpus, licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0). The VCTK Corpus was created by the Centre for Speech Technology Research (CSTR), University of Edinburgh.

The CCS10 (English) portion of the Mozilla Common Voice dataset, licensed under Creative Commons Attribution 4.0 International License (CC BY 4.0). Mozilla Common Voice is a community-driven dataset contributed by global volunteers.