User blog comment:CykeP/Mixing Help - Please Read!/@comment-53539-20160408081739

It goes back to the whole "right song, right Vocaloid" idea. Clarity is not always the best attribute a Vocaloid can give you.

I can't speak for the Kagamines from usage, since I have NO PLANS to buy them -ever- (I think their quality is **** and there are better vocaloids for English to do the same job, so I'd rather get a better vocal over them but thts me, I've never been a Kagamine fan so I'm bias against them as a fan). So I'll talk from my worst vocaloid experience - Sonika. My least favourite English vocal (due to REASONS), one way or another. Seriously, I don't like Megurine Luka English, Cyber Diva or Ruby.... But Sonika bets them all as a English vocal.

One of the things that used to help Sonika, is layering the vocal. I found if I added Big Al, but lowered the vocal so Sonika was the most notable vocal (very hard, V2 Big Al dominates most other V2 vocals!) that some sounds became slightly more clearer. I used it because I didn't ness. want to loose the traits within the vocal (like Sonika's softess), I just wanted her to be a *little* clear. Prima worked too, and due to being a closer female tone then Big Al, I could go a little louder. It goes back to the classic V2 technique of mxing Sonika and Luka... You should be able to do this because LEn and Rin are the same vocal with a different pitch (lets just get this out) basically with little difference between them. The draw is that it then can't be contributed to "Sonika" souly and there is a risk of the vocal being used to enhance the other not matching up. Saying that, even layering several copies of Sonika singing was at times good enough to improve the vocal slightly. So another trick I did with her was to have 1 octave above and below slightly mixed in. These are simple techniques, but they aren't miracle workings. So I'd not rely on them too much, its just food for thought. A lot of the time, experimenting with reverb, viberto and vocal effects in a good DAW produce much better results.

Note that some techniques with adding clarity and removing that muddy effect have hidden draws backs. The trouble is, if the original sound quality is not good that sometimes when you sharpen sounds or attempt to harden consonants, you get other draw backs like digital noise and other flaws becoming sharp too. It something to be aware of, because there are members of the fandom who will indeed give you good advice on how to sharpen the vocal and it often works well enough... However, all too often they IGNORE their also enhancing these bad results because they don't notice them. This was an issue I had with Circus-P in the early days (I think) because I couldn't stand his rendering of Sonika because it really brought out her clarity AND her noise loop issues.

CFM English vocals are difficult... They often sound forced to me, even with the best producer attempting to use them entirely in a song... Its either because they have to voice act the vocal and do a language their not familiar with, or best their done on the cheap. Which is why they basically have a sucky 1 octave English vocal range, which any other English vocal basically beats. No kidding... This is the SAME issue that haunts Gachapoid V2. And just as it effected his quality, it lets them down. =_=

In short, the vocals are best for that "1 line of English" Japanese songs often have randomly and are not as good as they could be. In addition, the Kagamines were in Japanese the "power" type of the Character Vocal Series and this "power" is not always present in the vocals. This may be intentional or not (Wat claimed this was with Hatsune Miku English, but I honestly do not believe him since I think it has more to do with the voiceress being unable to produce good English results, and the fact he is working on another English vocal). But practically ever vocal in English has issues matching their Japanese (Kaito sounds like his singing through his teeth for example).

As someone said, you can often swap sounds for English vocals (with the exception of Luka V2, due to missing sounds). There are about 20 vowel sounds in English so switching vowel sounds often works. Though there is a limitation here like with Luka V2. Its something people on't like about Macne Nana, because the original vocalist was Japanese sometimes Japanese -> English vocals result in less variation between vowels. It varies greatly... In the case of Nana it was intentional. In the case of CFM its not... This is where the "forced" aspect can come into play, as its a common thing for tension on sound to be wrong they may stress a sound that is in English more relaxed, or not put enough stress bead another that needs it. So when you work on the Kagamines, watch out for this too in case it pops up.

English vocals are not easy to work with, and while the Kagamines are not the worst... I wouldn't say you've gone with the best choice. In addition, as their still less then a year old, some producers may be still discovering how they work (though a lot of things will be known by now).

Also... Sigh... Bare in mind, people think Vocaloids a really good engine... But some of the clarity issues, particularly muddy sounds... Are just Vocaloid being Vocaloid. You can improve the vocal all you want, but Vocaloid effects how a vocal sounds. Off the top of my head, Ruby is confirmed to be the case. Her provider Misha confine the vocal samples behind her are not nasally, so Vocaloid is impacting the vocal here itself. Its another issue some just don't want to Admit about Vocaloid... This was nothing new too - Gackpoid has the SAME issue about his vocal. =_=