Retrieval-based-Voice-Conversion-WebUI

Retrieval-based-Voice-Conversion-WebUI copied to clipboard

[go-web.zip](https://github.com/user-attachments/files/22582512/go-web.zip) - install 'vs_BuildTools.exe' and 'python-3.10.11-amd64.exe' ( ✅ MSVC v143 ✅ Windows 11 SDK ✅ C++ CMake tools for Windows ) 'install_rvc.bat' ``` @echo off set PYTHON=C:\Python310\python.exe echo [*] Downgrade...

Khampol

The fairseq code error RVC V2 Disconnected

1

 Command '['pip', 'install', '--upgrade', 'fairseq']' returned non-zero exit status 1. I already to send message gmail Kit Lemonfoot noting can fix this notebook.

cahayamahardika16

I have a pth and index file I downloaded. I am trying to convert these to an onnx file using the webgui. I'm not exactly sure what I need to...

YashinNeko

I Use it on Ubuntu Real-time GUI can't see my pth files in /weights as well as index files If I text files names manually and start VC, GUI just...

nikitapogromsky

Audio training

1

Hi, how are you? I want to obtain the model files (.pth and .index) trained on the audio file I want to replicate. Please confirm if this application supports that...

smart01250

Hello is Multi speaker training supported? if it's not possible to do it from the web ui, is it possible to do it from the terminal? Any tips would be...

mhciwnzoq

I have completed training a model on RVC which works but would know like to remove unwanted files. The training files I used have already removed but would now want...

AkterAhsan

Batch size 4 vs 8

2

hi I have a 3050 6gb,so the best Batch size i can work with is 4 but i want to know how much time would be saved if it was...

itsthepatrick

Why does rvc can't handle emotional voice

7

Rvc can't properly clone crying,shooting , screaming like voices.., Is there any possible way to solve this issue

Chiyan200

H800训练会使loss异常增加到nan？

3

不知道有没有人用H800训练过。 我的个人电脑，windows系统，显卡3060，python3.10, cuda11.7,torch2.3可以正常训练，20几个epoch就可以将初始为31的loss_mel训练到15左右，得到的pth和index推理结果也正常。 计算服务器是linux系统，显卡H800，python3.9 3.10, cuda12.2, torch2.1~2.5都试过，使用1-4块H800都试过，使用和前述同样的数据和配置（基本就是gitclone下来之后没再动过），但训练时loss_mel会从30逐渐升至50，60，之后变为nan。得到的pth和index推理出的结果几乎为纯蜂鸣声。 看了其他问题的回答修改过fp16_run 为false，不起作用；减小learning_rate的初值，虽然在200轮以内不会出现nan，但得到的pth推理出的音频仍然充满电流声与蜂鸣声。 不知道是cuda版本的问题，pytorch版本的问题，还是显卡的问题？有没有用H800出现同样问题的？

jphtd