UTAU Tips

Japanese Pronunciation

I've found some people have a hard time making their Japanese UTAU sound native, so here's some tips I've picked up by listening to native UTAU's samples!

1. Vowels: Unlike English, Japanese has no relaxed vowel sounds, meaning 'a' will always sound like 'aah', 'i' will always sound like 'eee' and so forth. Make sure you aren't saying 'kuh' instead of 'ka'.

2. R: This is the trickiest Japanese phoneme to pronounce, as it both sounds like an English R and doesn't at the same time. Don't pronounce it like 'l', rather a more open-mouthed R like it's a vowel (I have no good way to describe it).

3. T: It's pronounced like the 't' in 'forty' (dx in the Arpabet).

Making Your UTAU Distinct

A lot of people when they're starting out just record the voicebank and then figure out the character. DON'T DO THIS!!

Think of the popular UTAU. Kasane Teto, Gekiyaku, Yokune Ruko, all of them. All their CVs went into recording knowing what their character was, and that's why they're so recognizable. Kasane Teto's voicebank was even pitch shifted afterward to give it more character.

In the end, even if you have a gajillion pitches, a 1000-dollar recording setup and an ungodly amount of appends, if your voicebank has no character it'll feel lifeless.

I'm not talking about 'character' as in 'mascot', I'm talking about what makes the voicebank recognizable. Yi Xi, the Synthesizer V voicebank, has no character attached but is still incredibly recognizable.

Before you start recording, take a few notes as to what you want your character to be like. Shy, angry, cute, you can even model them after a Vocaloid. When I was recording PIERO, my second voicebank, my notes were 'cute, young voice'. With just that, he became so much more unique than my first voicebank, where I had no notes.