I feel sorry to be biased, but since all Voiceroid series (1/2/AI) have bad code quality (bugs and bad designs), I hope they could recreate an editor rather than reuse any UI.
Otherwise, I don't think it would be a strong contender for CeVIO AI.
But I'd admit the AI VOICE may still has some...
The emotion part is not that convincing since the 2 "mini" Kotonoha voices still don't have emotions ("deep learning calculates that"), and the voice sizes are not reduced much.
"Export to mp3" won't be called as a huge improvement since I can add such feature to Voiceroid2 without much effort...
Thanks for using my UniTranslator. And thanks for showing my name with that image.
It seems that you're also trying to translate VOICEROID2, which I have tried several years ago (I just made a basic translation for English, no time to continue).
@uncreepy has explained most things. You can change your system locale to Japanese to use Voiceroid EX.
But even if you are using Voiceroid 2 with Kiritan EX imported, if you don't set your system locale to Japanese, you still cannot use the voice in V2. My locale patch can help you in...
Just tried this. Works well as a free, open-source tool. Process a 4 min song in 20s with 6GB memory used (CPU mode).
In most cases the result is still a bit worse than some commercial tool's output. But it can still be improved if someone would train a better model.