Vocaloid Wiki


English Vocaloid + realism

Youknow... There is ONE ever lasting comment that seems not to wane nor become less as an opinion. Its the "realism" of the English vocals. I'm not going to give examples of "realism" though here, because if your not convinced, you never will at this point.

Realism is something of an issue thats been around since the very beginning and has brought into question about the Vocaloid engine and its capabilities. But for some reason, English vocaloid gets that "Microsoft Sam can do better" level of reaction towards realism. Yet... Right off the bat, since V2, before Miku even came out Sweet Ann was noted for the uncanny valley level of capabilities that were being given out by Vocaloid2. Then there is the classic "Prima and Tonio have been known to fool long time opera fans" example.

Finally... I decided to show a few people Oliver singing, without telling anyone its a computer software signing... They'd never heard of "Vocaloid" and thought Oliver was real until I explained her wasn't. I did this two years ago and the people were my mother, landlady and two friends of hers. They are all 55+ years old.  :-/

So... The way people react towards English vocaloids and "realism" is kinda off. WE know their not real so WE look out for the errors and the way they don't sound right. And in the age of autotune, honestly, it means your LESS likely to even bother looking out for fake vocals and more likely to presume everything with the slightest bit of machine-editing is just autotuning. Synthesizers, for those of a older generation such as my mother's generation, don't even consider synthesizers at times because synthesizers in their day just didn't do vocals and the few that did sounded really bad. They've never come across the "vocaloid" level of synthesizer, which CAN have that "uncanny valley" feeling.

Then we get onto the other languages... I remember when I first lurked in the fandom in 2009, Miku wasn't considered what she is today. Yes, we know she is unrealistic (though with good editing can sound realistic), but once... I saw her compared to English vocals as being more realistic. At that point the Japanese still considered English vocals more realistic because most of their vocals were like Miku. I mean, this point don't forget the number of vocaloids to them was only about 8 for Japan. Not a great selection and all but 4 were realistic; 2 were on V1.

Its taken a while to make people understand that English vocals are difficult to recreate because of how much variation there is in the English tongue. But sometimes... I think Vocaloid fans tend to over exaggerate things.

I bring this up because;, there is one comment there that says that English Vocaloids don't sound like real people and then they go onto say Japanese and Spanish can.

Regardless of what anyone says; they can do realism, whether or not you care to accept that much they CAN sound like real people.

After spending some time trying to work out why everyone jumps on the "they don't sound realistic" bandwagon, I've started working out half o the reasons why this is a presumption. Putting aside the fact I mentioned earlier that people if they know a vocaloid is a vocaloid they go in knowing their not real... Part of it is also down also to the fact they get these English capablities thing stuck in their head. Bad English capabilities =/= a sign of lack of realism. Its just as good/bad tuning do not equal bad/good vocaloid. Its all in the hands of the producer and producers learn work around when faced with a bad sound.

But here is the thing... If people can pick up that Puppycat "sounds like a Vocaloid" even if they don't know more then the CFM 6, then there is *something* familiar all vocaloid share. Its also a sign Oliver is on par with Miku if they can do so, so at the very least English vocaloid must be on par with a Vocaloid once held in 2009 as more realistic then a English vocaloid. You see where I'm going with this line of thought right...?

I don't know... I go on about English vocaloid a lot so everyone is tired of me saying this stuff. At this point I thought some of these things would be gone as more languages enter the realm of Vocaloid.

Its just every time this issue comes up, people will post examples of realism and they still only amount to Japanese, and now occasionally Spanish, as examples of "vocaloids sounding realistic". So Chinese and Korean still need work it seems.

Honestly... I don't know about this whole issue of "Realism" anyway.

This is like that "there has been no improvements to English vocaloids" comment I saw last year.... Yeah sure, English vocaloid itself still needs work even with Diva out. But to say this comment basically meant that English vocaloid never got any better then Leon and Lola. Just in V1 alone, there were improvements between them and Miriam. Aside from Sonika (who has CONSTANTLY been used as a example to run English vocaloid down since she was released), there isn't any bad vocals. Sure, many of the Japanese -> English vocals need work; English vocaloid will always face the accent problem and honestly, non-natives will always prove to be the hardest to work with.

But to say English Vocaloid hasn't improved at all is basically Vocaloid itself hasn't improved since 2004, which as we know is BS because every Vocaloid doesn't sound like a V1. So to say "English vocaloids can't do realism", is basically saying the Core engine that powers English, magically can't recreate the human voice that allows it to make Spanish or Japanese vocaloids magically "realistic". If we imagine for a moment that every vocaloid language so far was equally on par with each other, they are once again saying "Vocaloid can't do realism". Yeah, it can't... But isn't this the whole issue with Vocaloid???

Its about being "uncanny valley" levels of realism.

Its also why extra vocals like the appends exist (once again) because every single vocaloid can only create a monotone level of realism. You have to do a lot of editing (and is the point of why XSY and vocalolistener exist) before you get anything that could pass for a real singer. But so long as the listener knows thats a vocaloid and the "noise" of the engine exists, this has points of which the logic will always be flawed.

Long story short the message I'm trying to say is;

Every vocaloid is capable of bloody being realistic - just because you can't tell when a Japanese nor Spanish vocaloid is not being realistic don't presume that the other 3 languages (English, Korean and Chinese) can't sound realistic at all.

This includes English vocaloids!

Also on Fandom

Random Wiki