I tried to use the Neutrino (AI/Neural vocal synth) and the model voicebank Kiritan.
This is an old original song composed by me and stella.
You really don't need to tune it too much, just place the notes, add lyrics and the AI will sing it but not realtime, it is offline render.
There are some limitations at the moment, like it can read only japanese characters not romaji, you can't add manually vibrato or command the AI to do that, you need 3rd party software.
I changed the pitch of the song to +2, i just want to make it a bit higher to sing.
