*UPDATE 2024: A.I.CHI's Diffsinger model has been greatly improved and it also is up for download! Check him out here!* • 【ENG + JP Diffsinger Release】Demo Ree...
Hi!! I got A.I.CHI's Diffsinger model to work with English using the ENUNUX Eng. phonemizer :-) !! I decided to import auto-tuning from Synth V for this cover, I just adjusted a few things and moved around a lot of the phoneme timings for this LOL but I am making progress w/ his model..!
Currently he is only at 57k steps, and I do plan on training him more myself in the future, so his quality will definitely increase more in future demos!
I also imported his dictionary so he can sing Japanese and English in the same track using the same phonemizer too, which makes cover making a lot easier !!!
I hope everyone also doesn't mind me re-using A.I.CHI art that i've drawn before.... I have been swamped with a full-time job and I am way too excited to post demos so I don't feel like drawing something specific LOL
--
Model Information
No. of Songs: 21
Sample Data: ~66 Minutes
Label Count: 22,133
No. of Steps Trained: 57,000
Model Production: PixPrucer
DiffSinger Port: haru0l
Recorded Range: G2 ~ C#5
Diffsinger: github.com/MoonInTheRiver/DiffSinger
OpenUTAU for Diffsinger: github.com/xunmengshe/OpenUtau
--
SVP by ChitoseP
Original - Mitski • Mitski - Washing Machine Heart
peeslubn
コメント