Preface#
Changed the title
Although it's said to be a rescue, in fact, I'm a bit of a clickbait
Because I really can't think of any words
In fact, the steps are still UVR processing accompaniment separation and reverberation removal RVC processing
But I think most of the problems still occur in the separation, noise, muffled sound, mostly because of this issue
The cracking and distortion are due to the significant difference between training data and generation, which is very difficult to solve, even after training several times it's still the same
And using this software directly, encountering noise issues cannot be fixed at all
Harvest restores the original sound more, but the effect will be poor if the singer's voice is significantly different
RMVPE, on the contrary, although the singing style is close to the input audio, there won't be any electronic pitch problems even with significant differences in the singer's voice
The software does not support AMD graphics cards, not at all, only CPU
Installation#
Visit https://www.tryreplay.io/ to download, then just follow the next steps
After installation, you have to wait a long time to download the model
Usage#
After installation, it looks like this by default
You can download the model on https://weights.gg/
Then
You need to import the model, select
Import both .pth and .index files
If it's trained with RVC, the pth and index are not in the same folder, you need to right-click on the model
Then copy them in, and rename both files to be the same
Configuration Section#
The following configuration will yield good results
If encountering electronic pitch problems, F0 can be changed to harvest, but it may result in voice muting and distortion
Conclusion#
It can only be said that it is a very convenient program
No need to worry about opening two programs and then crashing the graphics memory
The problem is, it cannot fix the sound