Ai Hoshino_(U)(Oshi No Ko)| Jp, En, Ru

First of all..... pain 私は気にしない。 私のお気に入りはこれまでもこれからも愛です Aqua: I don't care. My favorite always was and always will be Ai Всё равно. Моей любимой была, есть и всегда будет Аи и только Аи Trained on rmvpe. About 1.7 hour dataset from 1 season of anime, interview, view point B 5000 epochs, Sl510000 Creators: <@697759543492608040>, @ganbareshamiko helper Aleks29 Speech to speech, txt to speech, can sing Works best on rmvpe Credits are much appreciated https://huggingface.co/MUSTAR/Hoshino_Ai_U/resolve/main/Hoshino_Ai_U.zip For more models from me https://www.weights.gg/user/cln3lyorr0009lc0802mqz7ww

ватафак а примеры где

Wait a little <:AiThumbsUp:846980392497709066>

https://cdn.discordapp.com/attachments/1098732404128092200/1157641933385117736/kotik.gif?ex=65195990&is=65180810&hm=0e4237dcb706f589c9ffe555518dd6435ca5fca731b89ca46778ab6bee760321&

Mephisto - Ai Hoshino Cover

*5000 epochs?!*

why 5000

Ultra model. Also merged dubs Jp90 eng5 ru5

ok lol

1.7 hour and 31mb index?

can you show loss/g/total?

Do u mean index logs?

https://tenor.com/view/oshi-no-ko-gif-10708136726141321141

Yeah, interesting tho. That on my previous model was about 0.5 gb index

this one in tensorboard

yea same thats why I'm so curious

Hmm, actually a do not have tensorboard. <:KanaSad:816156368268034079> I'll do it tomorrow

if its not too much trouble please Im very curious

No problem. I'll do it tomorrow and pinn a screenshots

can you send an example without the music, please? i really want to hear how it sounds on it's own

got some myself

https://cdn.discordapp.com/emojis/1147604764448784405.webp?size=48&name=prayge&quality=lossless

thanks. it's indeed amazing

heres some samples with okada vc

слушайте, эта модель прекрасна хоть у меня с ней только один кавер xD