How do you get multiple characters from one sample?

PieCrust

Momo's Minion
I'm working on making my own voicebank at the moment, and I noticed in some VBs, and I noticed that in some of them, instead of having one recording for each character, they use one recording for a group of characters

Here's an example from Ana Nyui's Lemon VB
Screen Shot 2023-01-09 at 7.09.40 PM.png

I was wondering how I could do this myself!

For further information, I have every sample recorded. I use a mac, so utau looks a bit different on my end, and I can't use oremo if that's required (I've tried downloading it several times to no avail)
This isn't necessary for me to make the bank, I'd just like to keep things organized if possible. Thanks in advance!
 

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
It looks like this is a VCV voicebank. The reason each recording has multiple syllables is to get every combination of vowel, consonant, vowel. So instead of recording one "ra" you'll have "ara" "ira" "ura" etc. You can record all of those as separate files if you want to, but most people prefer to have them combined together so it's faster to record.

To keep track of everything you need to record, you'll want a prewritten japanese VCV recording list, or reclist. You can just read the reclist as a plain list, or print it out and cross out each line you record, or load it into a dedicated voicebank recording software to automatically name each file.
There's an official Mac version of oremo here, but I am not sure how well it works on the latest Macs.

You can get a reclist and a base oto from this website.

In the oto.ini there are multiple configuration lines for a single WAV file. Each of the lines has a different alias, and different timing parameters so that they line up with the correct syllable. You type the alias into the note lyrics to pick which OTO you want to use, and don't use the original file name at all.

Since you're on Mac, you're either using UTAU-Synth or OpenUtau. Both of those have automatic support for VCV, so you only have to type the lyrics as plain hiragana CV and the software will automatically insert the starting vowel for each note.
 

Similar threads