Dec 16, 2023
seen from Belarus
seen from Malaysia
seen from Italy
seen from Uzbekistan
seen from Israel
seen from Australia
seen from China
seen from China
seen from United Kingdom
seen from Mexico
seen from China
seen from United States
seen from United States
seen from Malaysia
seen from China
seen from Malaysia
seen from South Korea
seen from United States
seen from United States
seen from United States
Demo for Berry, a new UTAU & DIFF-SVC with a childish voice. :3 Her UTAU voice is bipitch and very versatile.
[ Download ]
*Art by julieraptor
Part II of V/V day!
The Gift From the Prince Who Brought Sleep PV but now it’s with diff-svc Christopher singing it.
(Pretrain with Nehito by MLo7, UST by Neemiso, rendered with Ryo by Aster)
RVC model for ZETHIEL_LV0 is here!! He's the AI voice model of my pre-T voice, so treat him well.
Input: Mai Synthesizer V UST: www.nicovideo.jp/watch/sm39221970
Use ZETHIEL ZERO AI RVC here (for RVC v2 and Kits.AI): zethiel.carrd.co/#third
Well, V day is approaching us. Time to kick it off with something wonderful.
In @thesilverlock ‘s paraphrased words, it’s like hearing a drunk Christopher “get really into karaoke” with a reluctant Kaito dragged on stage
anyway, since Twitter is burning, here’s the latest fruit of my loins
sorry no art and instead im having an autism vocal synthesis enjoyer moment, but i wanted to post a diff-svc voice model ive been working on with my own vocals in between work stuff and whenever i have the time to record and to allow my pc to be on to train the model. like 11k steps so not very high in quality but its insane how accurate it is to my own voice already. info doc regarding diff-svc for the uninitiated, but think the vocalo changer from vocaloid 6 but free* and you can train your own vocal samples to sing or say literally whatever from whatever language.
credits for the svp/synthv file i used as reference audio for this sample using my model in the vid
*i paid for google colab pro for more resources because i dont wanna train locally lol
Training Sonika AI morph and she still sounds like shit at 108K steps even with pretraining and over an hour of data she is simply living up to her brand to the end
