Nishiren's AI singing voice synthesis model for Diffsinger. They are Nishiren Gard in their future, where they have embraced their humanity more. Nishiren has gotten better at socializing and expressing themself from their career as a virtual singer.
AGE: 22
RELEASE DATE: 2/4/2024
LANGUAGE: English, Japanese
VOCAL MODES: Standard, Power, Sweet, Soft, 2P*
(*only to be depicted by Kayama)
DATASET: 40 mins main, 15 mins vocal modes
Public datasets used for multispeaker: Namine Ritsu, Ofuton-P, PJS, TSVD, Alex Floarea, TIGER
ART: Matcha_Worm
Quick showcase of Standard and 2P vocal modes using their 120423 alpha version of their model.
Introducing Nishiren diffsinger's experimental vocal mode- 2P (a.k.a. Kayama)!
— Gardanana (comms open) (@gardanana582) December 5, 2023
Made by labeling their cvvc bank and 20 secs of extra english data (~4 mins total). Also check out how they both sound with their finetuned vocoder! pic.twitter.com/tWDi1LBka7