XTTS-FemaleSerana / README.md
AIstDave's picture
Update README.md
f65c95e verified
|
raw
history blame
663 Bytes
metadata
language:
  - en
base_model:
  - coqui/XTTS-v2
pipeline_tag: text-to-speech
tags:
  - XTTS

Instructions: Just copy the files into your XTTS-WebUI main directory. Also I recommend that you disable DeepSpeed. While it does cut output times in half, it greatly reduces the output quality.

Version: 1.1.48 Pre-release. About this version: This model was built on a manually curated dataset. The dataset was initially created with whisper in step one of XTTS-Finetune. The clips were then manually edited to fix the issue of the clips being cut to short. Also the dataset's metadata was corrected for spelling errors. Dataset length: 425 clips, totaling 22:16.