tune can be messed with more

i cant find the right tune for me

Does the dataset of the voice sound like that? Because you only provided female ASMR voices in your pretrain though it surely does still sound appealing to me. Honestly maybe I should test it with this pretrain to see if it actually sounds as weird and odd as this one. I'm not really a quality person though so I have a different opinion and perspective with Ai voices.

yea this sounds like him (i just need to figure out the right tune for me)

ah

Okay I listen to a clip of his voice on YouTube and my god it sounds different from the RVC model itself. I think this might be difficult to get the tune like you gotta like make the voice half way whisper or have a quiet voice in front of your mic.

Well i used like none of his whispering

oh

whispering and rvc dont go well together at all

This is pretty much what all of the audio sounds like

like no whispering just him yapping

I have trained a RVC model that also has a deep voice but I used the Ov2 pretrain and it does sound as accurate to the original voice, this can worked in other pretrains too. I believe some pretrains are built to specific voices and some voices that sound very different tend to sound different or are a transformed version of the existing voices in the pretrain. It is his voice but modified within the pretrain hence he sounds different from the dataset It reminds me experiencing in playing around ElevenLabs or certain Ai tts voice were it sounds inaccurate to the actual voices but it uses the dataset as the reference to recreate the voice as it sticks with the catalog of voices they have their datasets.