Talk:Kagamine Rin/Len/@comment-70.26.134.243-20150831161937/@comment-86.145.196.46-20151003083355

Lets see...

In the V2 era it took 4 hours to record a single Japanese vocal, the major sounds in several pitches. The total number when completed was something like 1,200 sounds in total. For English you're looking at bumping that up to 2,500 per pitch and V2's had something like over 7,000+ sounds. It takes a week-1month to record it.

And thats without noting that V3 Japanese vocals had some 200+ era triphonetic sounds, as these didn't exist in V2 voices. And V4 has even more.

Some sounds can't be recorded and need to be made, so there is some of the development time taken making sounds from mashing two recorded sounds together. Or creating sounds like voiceless sounds (English, French, langauges like this, tho for the moment Vocaloid over covers English take note).

If you count vocals like Miku, the Kagamines, etc, then they had extra time taken to process their sounds, though vocals like the Zero-G ones (minus Sonika) didn't.

Thats without having to quality check each of those hundreds/thousands of transistions per sound (something like 10,000+ for English) that go into word formation. Tweak sounds that don't sound right... Etc... Etc... All while trying not to loose quality... . See Cul's problems, they had to re-make the vocal more then once from the recordings. Then there are vocaloids who have to sound like their providers which means making sure its got enough traits... Scheduling the providers alone.... Thats tough. Sometimes youneed more then 1 session.

Then you have to check in with Yamaha before release so they can make sure the vocal is top notch. And Yamaha has the final say.

This is why UTAU vocals at their HQ never match up to Vocaloid's HQ. Vocaloids have layers upon layers of sound in several pitches. The microphones used are generally better then most UTAU makers are using, the sounds are checked and re-checked... Even at its worst, the LQ vocals for Vocaloid are STILL classified as HQ vocals, otherwise Kagamine Rin/Len and Sonika would never have been allowed release. Quality does vary, still between vocaloids themselves. :-/

Its not Vocaloids can't be done quickly, its just it often notices when it happens too fast. The Kagamines being the prime example, to exploit the Vocaloid craze at its early stages were coughed out in a hurry... And they paid the price, being worst quality then Miku, who herself by the end of the V2 era alone was reduced to LQ because there were HQ vocals released. :-/

Bare in mind, most comparisons of Vocaloid versus UTAU don't take the time it takes to make a vocaloid into account, the depth and scale of the projects, etc, etc. As its been said, there is a little more then autotuning going on here. The more experienced the developer though, the FASTER they work, since they already know what works. Though Vocaloid voicebanks have a bit of trial and error behind them regardless.

All of this, amounts up to way it takes so long to make a Vocaloid. If it didn't, the product itself would be effected. Also, this is why a Vocalodi engine upgrade happens once every 4+ years or so, to give time to make Vocaloids. Each company has several planned or on going and each engine release results in a project being reset from scratch. So Yamaha can't produce a new version every year. and thats without considering some Vocaloids now come with several expansions on their vocals like Megpoid, Miku, VY1, etc. Would you prefer a crappy quality for the sake of speed? I'd rather a Avanna then a Sonika...

Off topic...

Autotuning, if used, adds eletronic noise to samples and I don't recommend it for raw vocals powering a UTAU. If you can't sing, then thats fine, but auto-tuning prevents the voices sounding as realistic as they could. Vocaloid doesn't auto-tune its samples which is why when Miku says "ma" the pitch and note aren't snapping into perfection and you may need to auto tune her to get there. In fact... Bare in mind that most Vocaloids ARE off-pitch and out of tune. Its pretty much long established, especially those from providers with little experience in music like Miku herself. Don't presume the Vocaloids are 100% perfect or getting 100% perfection is the ideal way to go. It lets down many of the UTAU that use it, as it makes them even more metalillic then they should be. Though if you can't sing... I can see why you'd use it... But consider its better to auto tune your UTAU's results AFTER you make it sing, rather then to MAKE it sing firstly.

Not that I'm an expert in UTAU here, but you should know enough about singing really before recording one... If not, then the UTAU is already flawed to begin with and I suggest you re-record it when you can do better... Its benefitcal to your UTAU... You get to do Act2, appends, etc this way and learn a lot more. :-/