https://github.com/vtuber-plan/flowvae

https://github.com/vtuber-plan/flowvae

pytorch vc voice-conversion

Last synced: 6 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/vtuber-plan/flowvae
• Owner: vtuber-plan
• Created: 2022-12-08T12:37:15.000Z (almost 3 years ago)
• Default Branch: e2e
• Last Pushed: 2023-12-12T07:15:28.000Z (almost 2 years ago)
• Last Synced: 2025-03-24T17:11:06.161Z (7 months ago)
• Topics: pytorch, vc, voice-conversion
• Language: Jupyter Notebook
• Homepage: https://blog.frostmiku.com/Flow-VAE-VC/
• Size: 1.46 MB
• Stars: 13
• Watchers: 5
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: readme.md

Awesome Lists containing this project

README

          
Unofficial implementation of FlowVAE

**code is updated after original method, pls check commits**

![arch](flowvae.png)

## Setup

1. resample the dataset sr to 16k

2. run `python 1.embedding_sid.py /path/to/dataset data/spkerEmbed` to extract sid

3. set train/val labels into the label dir, format as `wav_path|sid_path`

4. run `python train.py -m YourModelName` to trian a vc model

## VC

1. modify `configs/vc_rule.yaml`

2. run `python 2.vc /path/to/sid /path/to/output_dir`