UTAU voice banks won't make certain sounds

[CarvedHeart:3]

I'm VERY very new to vocaloid production, and I've had an oddly specific problem regarding UTAU that I can't find a solution to anywhere :/

I'm using a Kasane Teto voice bank on whatever the current version of UTAU (not OpenUTAU) is, and for some reason certain lyrics just won't play when pasted in. For example, I'll paste in 雨雲の灯りで瞳をつないで, put it in, and it won't play 雨雲, 灯, and 瞳, but everything else will play just fine. If it has to do with those characters just not being in the voice bank, is there any way to at least mimic those characters?

 

[sunnyp4rk]

Are you pasting in the kanji? It has to be in hiragana. That's how the majority of JP UTAUs are configured.

 

[CarvedHeart:3]

Are you pasting in the kanji? It has to be in hiragana. That's how the majority of JP UTAUs are configured.

Yeah, that’s how I’m doing it. Is there a way to covert it or do I just have to figure it out myself?

 

[sunnyp4rk]

Just look at the romaji of the lyrics (if there's not a hiragana version on the lyrics page) and then put the respective hiragana. I don't really know a better way of doing it tbh.

 

[MagicalMiku]

You can use the japanese dictionary Jisho:
https://jisho.org/
Just paste the kanji, and in the results you'll see both the meaning and the hiragana words (on top of the kanji).
For example, if you search "雨雲", you'll get the hiragana words "あまぐも".
For more text, you can use Google Translate, you'll see romaji in that case.

 

[Rylitah]

To explain why this is the case, UTAU is an old program made by a single developer (Ameya) - it doesn't have the capability to know what lyrics you're inputting when you type in kanji since kanji can have different readings/pronunciations and mean different words depending on the context (which the program can't understand).

When you input a note, you're essentially just telling the program to play the specific recording that corresponds with the sound you write down at that time, at that pitch. UTAU banks are comprised of recordings of a bunch of sounds and not words, so the notes you put down are you essentially splicing together a bunch of those recordings/sounds to make the words. This is also how Vocaloid works (at least pre-V6/AI), but that program has the capability to pull the sounds together to form whatever word you type in - not the case with UTAU.

You'll have to split the notes in half (or more) to properly fit all the sounds that make up the word you're trying to get Teto to say. It's a bunch of trial and error, especially since UTAU is so archaic at this point (just a fact, not necessarily a diss since many people still prefer working in it), but once you get the hang of it and get why it's like that it makes a lot more sense!

 

[CarvedHeart:3]

Thank you all for the help! I’ll use your advice and hopefully it’ll end up good! :3