Hey~! This page is informative~!
The content of this page may be of interest to readers, contributors and visitors.

This article in a nutshell states:

  • VCV/Triphone Voicebanks


Triphones are simply a different way of recording banks. There are two main methods, lite lists, such as the methods used by Suiga Sora, and standalone banks, used by Ritsu Namine and many others.

The Japanese term for this is 連続音, or "continous sound". Many users, both Japanese and overseas, refer to it as VCV (vowel-consonant-vowel.)


The first UTAU to use a VCV bank was Momo Momone in her song, "Kenka Wakare." The concept of a VCV was created by Ameya, UTAU's creator.

How It Works

VCV is usually recorded in strings of five or seven kana, such as ”かかきかく”. By configuring the bank in a certain way, you allow that sample to be divided into "- か”, ”a か”, ”a き”, ”i か”, and ”a く”.

Samples are configured by putting the overlap on the full part of the previous vowel and the preutterance between the consonant and vowel. Using UTAU's crossfade option, this creates an affect of a 'human transition'.


VCV voicebanks are without a doubt more realistic than CV (consonant vowel, standard) banks. They are often said to sound human. Certain plugins have been developed to make using VCV banks easier to use, such as converting a CV ust to VCV and allowing you to chose a different prefix.

Identifying a Triphonic Voicebank

It is very rare to see a triphone bank that are not constructed by creating a list of strings of kana and recording from them. Recording triphones in two-mora samples is not reccomended, as UTAU can only handle less than 1000 samples without lagging. (Although this was fixed and it supports more than 30,000 samples without lagging)

With a good oto.ini and a clean recording, triphonic banks tend to sound more human than their CV counterparts.

Lite-List VS Standalone voicebanks

Lite-lists are around twenty or thirty strings of kana that go along with a standard bank. It will sound more human in some places, though there is a noticable difference between the CV and VCV components.

Standalone banks seem to be more common. There are about three hundred strings of kana in a standalone list, and tend to include samples aimed at English compatibility, such as "すぃ/sui", which is a stand in for "si" due to Japanese replacing "si" with "shi".

VCV Voicebanks

See: UTAUloids with Triphonic Syllables

Community content is available under CC-BY-SA unless otherwise noted.