what reclist does vocaloid use

xyompltrine

Teto's Territory
Defender of Defoko
This is the most dumbest question I've ever asked but,
I'm wondering what reclist (Japanese) does vocaloid use.

When you edit the phoneme you can do /de s/,
Is it like a CVVC reclist or something else like that?
 

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
Japanese CVVC is probably what you're looking for, if you want to be able to put consonants at the end of syllables.
 

dead_byte

Momo's Minion
VOCALOID uses its own recording scripts, which are not available to the public.

VOCALOID1 & 2 only supported diphonic synthesis. Basically, it's something like CVVC in UTAU, though it's more similar to DeepVocal's CVVX format. That goes for both Japanese and English banks.

VOCALOID3 introduced support for triphones, as it was one of its most significant changes from V2. From that point on, most Japanese voicebanks seem to use a recording script that's probably similar to VCV, while also having VC-esque and standalone C configurations for when singular consonants are called for ( or possibly VOCALOID just calucates those by itself somehow, it's hard to say ). English banks on the other hand became a bit more complex. There's not necessarily a clear way English banks use triphones, especially since reportedly Cyber Diva ( who is on VOCALOID4 granted, but bear with me ) has around 230 triphones per pitch layer, which would seem quite low if the voicebank was taking advantage of triphones the way Japanese voicebanks do. Granted, triphones aren't only VCV, as they can exist in other configurations, such as CCV, VCC, CVC, etc. My guess is that they're mostly used for better consonant clusters, and perhaps some CVC combinations previous voicebanks struggled with, but I don't know that for certain. What I do know is that there is a general consensus that post-VOCALOID3 launch English banks were typically regarded as clearer than their VOCALOID2 counterparts. Then of course, you've got Spanish, Chinese, and Korean added in the V3 generation. I don't know how many triphones Spanish takes advantage of, it could very well be similar to a Japanese bank, since Spanish is fairly easy to synthesize; in addition, I would gather that the glides present in Spanish voicebanks probably take advantage of triphones as well, thus is likely part of why Spanish voicebanks weren't implemented until VOCALOID3. For Chinese banks, I'm like, 80% sure that they're basically multipitch VCV Chinese banks ( excluding Miku V4C, who is probably diphonic based on how much choppier she sounds compared to other Chinese VOCALOIDs. Or maybe she was constructed with artifically made triphones that didn't turn out so good, idk ), thus is probably why they seem to take forever to develop. As for Korean... I'm not gonna pretend like I know how Korean banks use them, if they even use them at all. To an extent they may be use like Japanese banks for VCV transitions, though I'm not sure how things like ending consonants are handled in relation to triphones.

VOCALOID4 and 5 are pretty much the same way. There have been revisions of the recording scripts to improve results, though they seem to use mostly the same synthesis methods.
 
Last edited:
  • Like
Reactions: xyompltrine

Similar threads