|
--- |
|
title: GPT SoVITS Template 0.2.0 |
|
emoji: 📈 |
|
colorFrom: indigo |
|
colorTo: purple |
|
sdk: gradio |
|
sdk_version: 4.16.0 |
|
app_file: app.py |
|
pinned: false |
|
license: apache-2.0 |
|
--- |
|
|
|
This space is designed for the convenient deployment of the GPT-SoVITS inference application. |
|
|
|
Step 1: Clone and Set Up the Repository Locally |
|
|
|
Duplicate this repository and clone it to your local machine (make sure to use git-lfs). Ensure that you have write permissions for the repository. |
|
|
|
Step 2: Organize Files and Weights |
|
|
|
Create 2 folders ('GPT_weights' & 'SoVITS_weights') to hold the weight files for both GPT and SoVITS. |
|
Place the GPT_weights and SoVITS_weights files in the folders. These files contain the necessary model weights. |
|
|
|
Step 3: Prepare Audio Data |
|
|
|
Inside the slicer_opt directory, locate audio files that are shorter than 10 seconds and longer than 3. |
|
Create a folder ('audio') as needed and move the selected audio files and slicer_opt.list file there. |
|
Modify the paths in the slicer_opt.list file (located in the audio folder) to be relative to the audio directory. |
|
|
|
Step 4: Commit Changes |
|
|
|
Use the standard Git workflow: git add, git commit, and git push. |
|
|
|
|
|
本空间用于便捷地部署GPT-SoVITS的推理应用。 |
|
|
|
step1: 复制本空间，并克隆本空间到本地(需使用git-lfs)，需确保git拥有write权限 |
|
|
|
step2: 创建两个目录并命名为GPT_weights与SoVITS_weights，将GPT_weights与SoVITS_weights对应的权重文件放入对应文件夹。 |
|
|
|
step3: slicer_opt目录下短于10s长于3s的录音文件和相应的slicer_opt.list文件都放入audio文件夹，slicer_opt.list的路径需修改为相对audio目录的路径（只保留wav文件名） |
|
|
|
step4: git add commit push三步曲 |
|
|
|
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference |
|
|
