Our goal is that the information on VocaDB is both accurate and reliable. Because VocaDB is a database specifically made for singing synthesizers, those synthesizer voicebanks are given more attention than on other music databases. If it’s known that the song uses a specific variation of a voicebank (such as Miku V4X Sweet), we want to display that information whenever possible. The problem is that often the exact voicebank isn’t mentioned anywhere officially. The artist might say that the voicebank used is “Hatsune Miku”, or possibly “Hatsune Miku V4X”, but you can clearly hear that it’s the English voicebank. Even worse, the artist might not specify the voicebank at all. Obviously this is a problem that has existed since the Crypton Appends were created for Vocaloid 2, but it’s only getting more complicated with V2, V2 Append, V3 and V4X versions of Miku out there. So what to do?

The safest way would be to go with what the artist said: if the artist states that the song is using “Hatsune Miku”, then you’d credit the original Miku voicebank, even if it sounds like V4X. If the artist says that the voicebank is “Miku V4X”, you’d use the “unknown” voicebank for Miku V4X. Obviously this isn’t always what we want, because while reliable, the safest way isn’t the most accurate.

The key point here to understand is that if we want VocaDB to be treated as a reliable source, we want the information as verifiable as possible. Taken from Wikipedia’s guidelines:

In Wikipedia, verifiability means that anyone using the encyclopedia can check that the information comes from a reliable source. Wikipedia does not publish original research. Its content is determined by previously published information rather than the beliefs or experiences of its editors. Even if you’re sure something is true, it must be verifiable before you can add it.

We don’t want to be as strict about this as Wikipedia is, because as said, being too safe harms accuracy. This is especially true when it comes to crediting voicebanks on VocaDB, because very often objective verifiability isn’t possible.

Therefore, we allow editors to input unverified information when they have a strong reason to believe that the information from the official sources is lacking and they can with sufficient certainty say that they know better. This knowledge might be based on experience with dealing with the voicebanks, or earlier songs by the same artist. Or perhaps the artist has posted that they bought Miku V4X and plan to use that for all future songs. Still, there must be some source for this information: if you as an editor use your best judgement to input information (such as credit a voicebank) with no official source, then you yourself are the source.

Official, verifiable sources are generally provided as external links to artist’s website, links to online stores, or video streaming services such as YouTube or NicoNicoDouga. Secondary sources such as Wikipedia, vgmdb, Vocaloid Wiki are also accepted. If there is no verifiable source, meaning you as an editor are the source, then you’re asked to mention that in some way, so that others don’t need to look for that official source which doesn’t exist. You can do this in multiple ways: edit notes, entry description or comments.

It should also be noted that, unlike Wikipedia, we do allow original research in the sense that artists are allowed (even encouraged) to input their own information. All information inputted by the artist who created the song is automatically verified on VocaDB, with the assumption that the artist knows for sure which voicebanks they used. Therefore, verified artists don’t need to provide references or justification for the voicebanks they credit.

This rule only applies when the artist him/herself is editing the entry. If you got the information directly from the artist, then you can relay that information to VocaDB, but you’re expected to explain it.

Sometimes choosing the “safest” voicebank isn’t completely straightforward either: maybe the artist says that the song is using “Gakupo Power” voicebank, but you know there’s Gakupo Power variations for both Vocaloid3 and Vocaloid4 engines, so which one is it? The rule of thumb on VocaDB is to use the earliest voicebank, in this case Gakupo V3, and apply the “unspecified voicebank version” tag. The original Gakupo V2 voicebank would be acceptable as well.

If you want to be thorough, you can also ask the artist to specify the voicebanks they used, to get the official confirmation. We don’t expect anyone to do this for all songs, for the reason that there’s so many songs with uncertain voicebanks. While this extra effort is always appreciated, it’s never expected.

I should also mention that it’s never wrong to pick the “safest” voicebank, so whenever in doubt, you should do that instead of guessing. And don’t stress too much: no one expects you to know all the editing guidelines from the start.

I hope this clarifies my opinion on the subject, and that it doesn’t sound like we’re being unnecessarily strict. I want to avoid making editing on VocaDB bureaucratic or tedious. I feel that this policy of allowing some degree of “best judgement”, when it is explained, is a good compromise between accuracy, reliability and effort. There is nothing new about this policy – it just hasn’t been properly explained before, which is unfortunate because I feel some unnecessary misunderstandings could’ve been avoided. Finally, I would like to mention that, as always, I greatly appreciate the work done by so many of our editors to create the most comprehensive and reliable database of Vocaloid discography information.

TL;DR: if uncertain, or you want to get away with as little effort as possible, use the information provided by the artist, even if you suspect that the official information might not be accurate. If ambiguous, use the earliest voicebank that applies and apply the unspecified voicebank version tag. If you know what you’re doing, you can use your knowledge to provide more accurate information, but please mention it somehow when editing the entry.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.