Open Neovolter opened 8 months ago
Is this a general problem you have? I have noticed it, but only sporadically myself. I am not sure it is a bug in the implementation. Perhaps it is simply that the vocal separation models in the pre-processing part of the pipeline have trouble with certain songs?
previously, with default settings, there was a clear changed voice in the chorus, but now, for no reason at all, when trying to replace the voice in the same composition with the same voice (same rvc v2 model) as before, the original voice remains in the chorus and the modified one is barely audible against the background