Should I record extra sounds?

[HowlingWolf15]

I mean sounds like くぃ　くぁ stuff like that. I'm recording my first voicebank, so I'm wondering if I should or wait until I become more advanced with UTAU.

Also a side question, but sounds like きゃ and related, when I oto do I cover the y over with pink or just cover it a bit like with a vowel

 

[Nat]

I mean sounds like くぃ　くぁ stuff like that. I'm recording my first voicebank, so I'm wondering if I should or wait until I become more advanced with UTAU.

Kwa, kwi, etc aren't needed in japanese (I don't think?) so you don't have to record those.
I can't really help with the otoing because I'm not that great at it (or at explaining)

 

[bio]

I say go forit, if you want people to immitate engrish or other languages

 

[MillyAqualine]

I'd say only if you wanna use them (and if you do like, half of the time at least)

 

[kamonohashi]

I mean sounds like くぃ　くぁ stuff like that. I'm recording my first voicebank, so I'm wondering if I should or wait until I become more advanced with UTAU.

Also a side question, but sounds like きゃ and related, when I oto do I cover the y over with pink or just cover it a bit like with a vowel

The sounds like くぃ and くぁ aren't needed. Any sample that has "-consonant- wa/wi/we/wu/wo" isn't needed. You could keep the ones that are like "-consonant- ya/yu/ye/yo" though, but they can easily be imitated. (like [ki] [ya] with the [ki] being very short.

 

[Terindie]

I have a resource on this! Should answer your question on what qualifies as an "extra". Use it to adjust exactly how far you want the voicebank's capabilities to be.

 

[Kiyoteru]

Side Question answer: You should use the spectrogram view to clearly see where the "y" is in "kya" and place the preutterance (red line) at the end of it. This is my personal preference, and other people will usually state their preferences for otoing these types of samples. You will have to experiment with the position of the preutterance for yourself to find what you think works best.

Main question answer: Yes, consider looking at terindie's resource. Personally, I believe the most useful extras are additional consonants. Things like "kwa" "kwi" can be recreated easily with "ku" "wa" etc. but there's no way to recreate a V or L unless you record it. These are the extra consonants I have in my personal CVVC japanese reclist.

Spoiler

l
v
th (unvoiced ie "think" "bath")
dh (voiced ie "that" "bathe")
ng (like "ing" but at the beginning) (alternate pronunciation of g)
zh (like "s" in "pleasure" "asia") (alternate pronunciation of j)
dz (alternate pronunciation of z)
r_ (extra r) (I use this one for rolled r) (written as "ra_" "ri_" etc)
r- (extra r) (I use this one for english r) (written as "ra-" "ri-" etc)
kk (aspirated/english k, which is pronounced forcefully with more air than japanese k)
tt (aspirated/english t, which is pronounced forcefully with more air than japanese t)
pp (aspirated/english p, which is pronounced forcefully with more air than japanese p)

Generally, if you get to the point that you want to add more vowels, then you're not really making just a japanese bank anymore. However, if you want to add some anyway, I would suggest recording only the V samples and no CVs. Here are the main english vowels that the japanese vowel system doesn't cover.

Spoiler

To prevent filenames being overwritten, capital letters will be written with a plus to distinguish them from lowercase letters. (X becomes x+)

i+ (XSampa: I) (IPA: ɪ)
The short "i" sound (ex hit, win, lick)
{ (IPA: æ)
The "a" sound in cat, laugh, happy
u+ (XSampa: U) (IPA: ʊ)
The "oo" or "ou" or "u" in words like cook, should, pull
v+ (XSampa: V) (IPA: ʌ)
An "uh" sound (ex hum, rub, sung)
3 (IPA: ɜ)
An "er" sound (ex teacher, bird, sure)
a+ (XSampa: A) (IPA: ɑ)
A darker "ah" sound. English accents vary a lot around this area... so it's hard to explain with word examples.

This is your first voicebank. Try not to overwhelm yourself with adding too much. With a Japanese voicebank, you should mainly be focusing on bringing additional expressiveness to Japanese, rather than reaching into multilingual territory. However, due to the limited phonology of the language, adding extras can help extend the capabilities of the voicebank.

If you'd like to add extras pertaining to other languages, you should do research on the phonologies. A wiki page will be much more concise about the different sounds you need to have than trying to get an explanation from a language speaker with no extensive experience teaching it, or by listening to language samples and attempting to pick out the sounds.