Do we need CC?

[methynecros]

Hello! I have recently noticed a tendency to create reclists without using transitions between consonants (I mean languages in which they are present). For example, not so long ago, the CVC format began to gain popularity, it doesn't contain CCs. Here are some examples of these voicebanks:

PT-BR:
RUS:

I would like to know your opinion, do you think CC is necessary for a quality utau VB, or does nothing really change without them?

 

[Kiyoteru]

The argument for CC samples is the same as the argument for VCV, VV, or VC samples. While you could theoretically go without any of those and rely on simple audio crossfading, you're inevitably missing the organic sound of human pronunciation. In an ideal world, there's a recorded sample for any sequence of phonemes, of any number of phonemes. It's entirely up to you how much you want to sacrifice.

 

[HitCoder]

In the two samples provided I can hear the unnatural transition between the consonants due to the lack of CCs. I personally see them as mostly necessary, but obviously from a dev standpoint it takes a lot of work to 1) record 2) configure and 3) use in a ust. so I understand dropping them. But for better results I do genuinely feel like CC is necessary. A "full" or "polished" voicebank would probably require them, a personal for-fun one probably wouldn't.