how to make a base CVVC oto

JamesOvO

Retired User
Retired User
Defender of Defoko
I made a reclist, but it doesn’t have a base oto. I don’t know how to make a base oto for CVVC, by the way it’s not CVVC japanese, it’s CVVC english
 
Last edited:
  • Like
Reactions: Kiyoteru

dead_byte

Momo's Minion
So, I don’t know if you’re asking for just a general answer, or just Kiyoteru to answer, but if you’re looking for general answers, I can think of a couple of ways.

For one way, there’s a tool on DELTA’s website called “mkototemp”. It’s a program that can generate base oto files, and is mostly meant for CVVC. That’s usually what I use for most voicebanks I work on. There’s also del_alias and doub_alias tools too, for removing extraneous alias, and duplicating aliases to another name ( so just as an example, making a duplicate of お, and renaming that duplicate to を ) respectively.

That said, I will warn you, this is more of an advanced technique. There’s no interface, and using it mostly involves adjusting the mkototemp.ini files until stuff sticks, and it can be pretty hard to work with. I usually have to generate several oto.ini’s and tweak the mkototemp.ini and try again until I get a base I’m satisfied starting off with, because mkototemp is very finicky and can cause some weird stuff if you don’t have the mkototemp.ini fixed to your exact specifications. There are also very few to no English resources for it either, so you’re going to have to translate a lot of the instructions ( unless you can read Japanese ). That said, theoretically you can make it work with a voicebank of any language, and is often used with systems like DELTA CVVC ENG or CVVChinese, so assuming that you’re not doing anything too terribly crazy, you can probably make it work with enough willpower.

And if that’s not an option ( like with a DeepVocal voicebank for example ) usually, I just make a set of configurations from scratch for a single consonant or vowel ( so for example, making configurations for all k-related sounds like ka, ki, ku, ke, ko ), and then abuse the “Replace” function in Notepad++ to create entries for other aliases ( so making duplicates of the “k” configurations, and then replacing “k”’s in the duplicate sets with “t” for example ), and then make adjustments as needed. That said, that can also get kind of messy, since it’s very easy to make mistakes, especially if the phonemes in the .wav files aren’t the same as their aliases, so sometimes you might need to get creative with the Replace function, or even learn to use some basic RegEx, but it’s better than just making a whole CVVC base from scratch.
 
Last edited:
  • Like
Reactions: nneko

Similar threads