>>1517591I used a few minutes (10+) of a youtuber talking, with varying degrees of quality/background noise and ambients that I filtered as much as I could with AI tools.
I used a freely available instance of RVC, that you can run through your browser, cause the local installation isn't compatible with Win7.
The link is dead, so you'll have to find your own.
I do remember getting subpar results even with pre-made voicebanks, but I didn't have a way to know whether they were high quality in the first place, or if proper settings were my demise.
If you want to test the capabilities of such software, however, I strongly recommend this ponyfag tool:
https://haysay.ai/Pick some character and try it out, I was able to voice all female characters in this animation with my own voice thanks to it:
https://www.newgrounds.com/portal/view/947079The way to make your own voicebank, however, is still opaque to me. Hopefully this can at least show you some directions, sorry I can't be of much more help.
You can also look for voice generals on /g/ archives, they are way more sparse than image generation because the field sucks for now.
