VCCV or Arpasing

It really depends. If you want to record it faster, than arpasing is your way to go. If you want the ust to be easier tuned, go for VCCV, as arpasing can use 6 parts for one word, whereas VCCV usually never uses more than 3 parts. As for otoing, Moresampler can give you a very accurate base oto, and takes less time. If you want it to sound realistic, really it depends on the tuning, and on you. Some can make arpasing sound more realistic, some VCCV. I suggest you to try an Arpasing Voicebank and a VCCV Voicebank, to see what fits you best.

I don't think one's necessarily better than the other. They're just different.

The main difference between Cz's CVVC list and Kanru Hua's Arpasing list is that Cz's is just strings of phonemes (eg, stA_sta_stE_ste) whereas Kanru hua's is strings of (mostly) real words (eg, prom on moms)

So his Arpasing list may be a little easier to read and pronounce. Its also shorter. However CVVC has recordings for some CCV and VCC combinations, whereas Arpasing is strictly diphonic, and iirc CZ's CVVC list covers a lot more phoneme combinations than Arpasing does, which has holes in it here and there.

Also, qft:

Sors said:

If you want it to sound realistic, really it depends on the tuning, and on you. Some can make arpasing sound more realistic, some VCCV. I suggest you to try an Arpasing Voicebank and a VCCV Voicebank, to see what fits you best.

Click to expand...

It boils down to personal preference. To my ears, VCCV sounds better than arpasing but I know this isn't the case for everyone.

Arpasing is a lot more convenient than Chezzie's list thanks to all of the tools available.
Chezzie's method tends to sound forced and unnatural due to the odd patterns but it works.
But Kanru is currently updating Arpasing and the experiemental version sounds a lot better.

My only gripe about arpasing is how ridiculously hard it is to use - so many small notes.

I personally prefer more Arpasing because

-It's much shorter and effective to record than VCCV
-Fewer sounds to voice configure (oto.ini) overall because it's strictly diphone orientated
-Recording list is very easy to customize for own accent/needings
-With Kanru Hua's default reclist: duplicates of sounds!!! <= super handy
-If write a custom Arpasing reclist, no need to use time to writing base oto.ini from scratch as Moresampler can autogenerate one for Arpasing: that might not be perfect purely used as that but it's ok accurate (especially with k/t/p sounds adjustments are required). So, recording tempo, style etc are very free!
-I like Arpasing's encoding (aka. way to name each vowel/consonant sounds). Like... for me "ah/ae/eh" makes much more sense and it's more memorable than VCCV's encoding.

Because how VCCV is recorded and configured, it's super "universal" to use it. VCCV is made a way that UST that is made for one VCCV works well in another VCCV by just plug-and-play*. *unless it's super accented non-native vocal that requires manual phoneme editing When gets used to VCCV's way of encoding, it's "easier" to use because it needs lesser notes overall than Arpasing when used.

VCCV has 2 types of "transition VC" sound both sounds same TBH - and one VC that is for ending consonants, so it's 3 types of VC in total). VCCV has 3 types of CC, one that comes the beginning of the sound, one that is put middle and one that is added to make an ending sound. And then there are two types of CV where one is "-CV" just like in Japanese VCV and one CV that is used "middle". And then there's, of course, VV for vowels, so those sounds smooth. So, it is CVx2, VCx3, CCx3, and VVx1.

Arpasing in another hand has just following and everything just once: "CV", "VV", "VC", "- C", "C -" and "CC". VC in Arpasing is oto like "V C" in Japanese CVVC, so it works as "transition VC". In Arpasing, to make "transition VC" to make an ending consonant sound, "C -" is added (often C - can be dropped out as real English speakers often lefts ending consonant unsaid like in Korean). "- C" is used to make "CV" to be like "- CV" I personally don't like to use this unless "s/sh/f", so I often left this out.

I have tried to record VCCV in past and it felt so... unnatural and forceful to record especially in parts where's two consonants in a row. And what I have heard from many that have recorded VCCV that quite many (not all!) of their VB can have really forced accent that is very similar to another... But I have heard some awesome realistic VCCV UTAU such as Al!ce.16, Minarai Mona and Fuutari Makku as the examples.

Arpasing and VCCV are both great and has its own pros and cons. It really REALLY depends on the user which one they managed to record/tune better. I, personally, managed Arpasing better as being short, effective and word-based but I do understand if it isn't suitable for some people. For people whose mother tongue isn't English and they do not speak English at all, Arpasing can be too difficult tongue twist. For they, VCCV can be better/easier option.