Print

Print


Vocaloid 5 came out recently, and it's got a lot of new features that 
look like they'd be useful. I downloaded the basic set today, which 
comes with 4 new voicebanks. Amy and Chris sing in English, Kaori and 
Ken are Japanese. There's also a premium version with another 4 
voicebanks, but two of those are minor upgrades of Cyber Diva and Cyber 
Songman (which I already have), and the other two are the Japanese-only 
VY1 and VY2. I've been trying out Amy and Chris on some of my old 
translations, and I've put up a video with lyrics from "Sy žumi reev" 
("A Whole New World" from Disney's "Aladdin").

https://youtu.be/0mX0QAlHJMk

Their Tirėlat pronunciation is pretty decent, but not perfect, e.g. 
Chris's unaspirated [t] doesn't always sound at the beginning of a 
syllable. Most English voicebanks have problems with sequences of sounds 
that don't occur in English. Amy's [d] doesn't always cooperate when I'm 
trying to fake a trilled /r/, but I didn't think to check if she has a 
[4] phoneme until after I did the song. It turns out they both have [4], 
but not [h\] or [e@0], and the [4] works pretty well for fake trills.

The biggest changes are the addition of "attack effects" and "release 
effects", which affect the articulation of the voice sounds, and the 
inclusion of style presets that add effects like reverb or EQ. I've been 
experimenting with those, but haven't done much with them yet. The video 
demonstrates the a style preset "Mood Expression" and automatic 
insertion of breaths, which is pretty good for making the voices sound 
more realistic. I've also been experimenting with the attack and release 
effects that add a bit of simulated "vocal fry" to the start and end of 
notes, but it's hard to get that to sound good.