What I mean is a website that can clone a voice without humanizing it, I need the text to speech to sound like those minecraft villager sounds that sound like their voices were shifted into sounding like specific words instead of the ai smoothing it out and making it sound unfamiliar. I do not know the terminology for this method or anything
#Looking for an ai website with "raw" voice cloning
formal hornet
mystic lily
What I mean is a website that can clone a voice without humanizing it, I need th...
What you’re describing is basically minimal post-processing inference , you want the model to output the raw vocal character without smoothing algorithms softening it. A few options depending on your setup:
If you’re using RVC, lower your index rate (try 0.2–0.3) and avoid any built-in noise reduction passes those are what humanize the output. You also want to make sure your source audio isn’t being pre-processed before inference.
For that Minecraft villager texture specifically, the pitch artifacts you’re describing come from formant shifting tools like so-vits-svc let you adjust formant independently from pitch which gives you that unnatural but consistent character.
Happy to help you dial it in if you want to share more details about your setup 👋