【 Ogata Masato / 緒正人 】 レミングミング / lemming ming 【 I made a DiffSinger Voicebank 】

hello im back from 2 months without uploads please give me the addoration i now crave
anyway yeah i switched from nnsvs to diffsinger, i only really did it because google colab updated to python 3.10 and broke all the notebooks, but diffsinger ended up having the same problems. also the nnsvs notebook was fixed so it doesnt even matter.
i went through hell and back trying to figure diffsinger out so that was fun. thank you to everyone in the server for helping me when i didn't even know what anaconda was.
the model is trained from scratch with 30 minutes of English and 30 minutes of Japanese. i didn't train multisinger as context can handle the accent changes when the vowels are seperated, with the exeption of the r's, which i just replaced the japanese r with dx cause their the same phoneme.
i think the recorded range is like D2 to F5.
il release the model when i have an hour of both languages (also i kinda want to record some irish songs but i cant diferentiate between like 2 phonemes and it haaaarddd)
Music/Lyrics - Kairiki bear - • 【公式】 レミングミング／かいりきベア fe...
Illustration - のう - @nounoknown
ENUNUX im using untill i set the model up for varience training - Makku - • INTERGALACTIA - AI Cov...
UST - アトリ科ヒワ属のゲン(囀る藤の庭) - • 【トースター音源改造】レミングミング【UTA...
Tuning/model/mixing - me🥰🥰🥰🥰🥰🥰🥰🥰🥰🥰🥰🥰🥰🥰🥰

[@psychic_digit]

Sounds pretty cool :) you are forgiven

[@nobodyremix]

FOR WHAT 😭😭😭

[@digital-p2123]

It’s been two months and no answer it must’ve really bad😭💀

[@psychic_digit]

@@digital-p2123 yea no it was bad

[@threadsofsaffron]

BANGER !!!

[@tiggywoovit5955]

I like this very much. I think I can forgive you

[@hansmichea]

create tutorial

[@nobodyremix]

Pix has a tutorial that walks you through creating a model, you may want to join the discord server if you need clarification on anything. docs.google.com/document/d/1uMsepxbdUW65PfIWL1pt2OM6ZKa5ybTTJOpZ733Ht6s/edit?usp=drivesdk