Skip to content

Latest commit

 

History

History
30 lines (18 loc) · 786 Bytes

README.md

File metadata and controls

30 lines (18 loc) · 786 Bytes
Raw

VCVITS: VITS-based Voice conversion Model

This is a repository for a voice conversion model based on VITS.

Pre-requisites

  1. Python >= 3.8
  2. Clone this repository
  3. Install python requirements. Please refer requirements.txt
  4. Download FairSeq Hubert pretrained models：Download
  5. Download datasets

Parameters

Training

File List Generation

First generate the file list text file from the dataset folder. One folder for audio files of each speaker in the dataset folder. Run filelist.py to get filelist.txt

Then split the training and validation dataset, run split.py.

Config Settings

Inference

pass