SoftVC VITS Singing Voice Conversion
Go to file
Ναρουσέ·μ·γιουμεμί·Χινακάννα ed31c18c30
Update LICENSE
2023-06-14 00:19:11 +08:00
cluster upload code 2023-03-10 19:11:04 +09:00
configs Add files via upload 2023-03-10 19:43:07 +08:00
dataset_raw Add files via upload 2023-03-10 19:20:48 +09:00
fairseq_onnx Add files via upload 2023-03-10 19:46:14 +08:00
filelists Add files via upload 2023-03-10 19:20:48 +09:00
hubert upload code 2023-03-10 19:11:04 +09:00
inference Add files via upload 2023-03-10 19:43:07 +08:00
logs/44k Add files via upload 2023-03-10 19:20:48 +09:00
modules Add files via upload 2023-03-10 19:43:07 +08:00
onnxexport Add files via upload 2023-03-10 19:43:07 +08:00
raw Add files via upload 2023-03-10 19:20:48 +09:00
vdecoder upload code 2023-03-10 19:11:04 +09:00
.gitignore upload code 2023-03-10 19:11:04 +09:00
LICENSE Update LICENSE 2023-06-14 00:19:11 +08:00
README.md Update README.md 2023-03-15 15:43:39 +08:00
app.py upload code 2023-03-10 19:11:04 +09:00
data_utils.py Add files via upload 2023-03-10 19:37:06 +08:00
flask_api.py upload code 2023-03-10 19:11:04 +09:00
inference_main.py Add files via upload 2023-03-10 19:37:06 +08:00
models.py Add files via upload 2023-03-10 19:37:06 +08:00
onnx_export.py Add files via upload 2023-03-10 19:37:06 +08:00
preprocess_flist_config.py upload code 2023-03-10 19:11:04 +09:00
preprocess_hubert_f0.py upload code 2023-03-10 19:11:04 +09:00
requirements.txt Add files via upload 2023-03-10 19:37:06 +08:00
requirements_win.txt upload code 2023-03-10 19:11:04 +09:00
resample.py upload code 2023-03-10 19:11:04 +09:00
spec_gen.py upload code 2023-03-10 19:11:04 +09:00
train.py Add files via upload 2023-03-10 19:37:06 +08:00
utils.py Add files via upload 2023-03-10 19:37:06 +08:00

README.md

Onnx导出

使用 onnx_export.py

  • 新建文件夹:checkpoints 并打开
  • checkpoints文件夹中新建一个文件夹作为项目文件夹,文件夹名为你的项目名称,比如aziplayer
  • 将你的模型更名为model.pth,配置文件更名为config.json,并放置到刚才创建的aziplayer文件夹下
  • onnx_export.pypath = "NyaruTaffy""NyaruTaffy" 修改为你的项目名称,path = "aziplayer"
  • 运行 onnx_export.py
  • 等待执行完毕,在你的项目文件夹下会生成一个model.onnx,即为导出的模型

    Onnx模型支持的UI

  • 我去除了所有的训练用函数和一切复杂的转置一行都没有保留因为我认为只有去除了这些东西才知道你用的是Onnx
  • 注意Hubert Onnx模型请使用MoeSS提供的模型目前无法自行导出fairseq中Hubert有不少onnx不支持的算子和涉及到常量的东西在导出时会报错或者导出的模型输入输出shape和结果都有问题 Hubert4.0
  • Fairseq导出时请将fairseq_onnx里面的所有文件复制粘贴到你的fairseq安装路径记得备份然后取消掉hubert那一段的注释导出即可