There's a sense of pride I get when I figure out how to make Miku's Japanese voicebank sing a word in English.
I have her VX4 version (cuz that's the same version as my Rin and Len and I wanted them to have the same quality). And yeah, that voicebank collection does have an English version, but the sounds aren't as clear. I'd much rather her have a distinct Japanese accent than to have fuzzy vocals. And yeah yeah, editing her voice and all that. I get it, I've been doing that, but truthfully, working with the Japanese version is easier for me in a lot of ways.
For one, I started learning the Japanese vocal editing first, so I had a better understanding of how to manipulate the voice and to sing a rough approximation of English words, and the more I edit, the smoother the sounds become.
See, I also use synth V, which is how I got good(ish I'm still learning) at editing phonemes and working with vocal synths in general. Miku's English voicebank allows me to use her more like SynthV, where I can type in an entire word and it will separate on its own, and it's less work trying to make the words sound English, but you can't slide phonemes around in vocaloid/Piapro the same way you can in the SynthV editor. So if I type in a word, I can't slide around how long the vowel is as easily. There are ways! But it's harder for me to remember and to implement the way I want.
Which brings me to: Miku's Japanese voicebank is easier to use because by the very nature of it, I am forced to manually blend and block out each phoneme. For the word "I", I type in "a" in one note, and then overlap the "i" at the end, which results in a very smooth "I" (the English word). For "I'm" it's the same, except the second note's phonemes are "i m".
More complex words like "crashing" are still difficult. The current iteration of the song has her singing "crashing" as "kerashin" (keh-rah-shin). Stand alone consonant sounds are difficult. Hard consonants like T and K are difficult to blend with other consonants, and the "ing" sound is hard to get a strong pronunciation on the "g". But, slowly, I am figuring out work arounds.
Words that end in a soft consonant (like the "d" and "b" in "end" and "job") are difficult at the end of a phrase such as when there is a pause/break between the end of one and the beginning of the next. But by separating that sound to its own note and adding a subtle vowel sound (turning "end" into "ende" or "job" into "jobuh") the ending is less cut off. The way I determine which bowl to use is where I end my mouth's position when speaking the word on its own. Usually this is dictated by the last vowel sound to be used, but it depends on where in the song it is and how harsh the sound is. Like, I rarely end a word like that with an "ah" sound because the "ah" is more pronounced than the more subtle "eh".
I'm far from an expert mind you, but I have found that a lot of tutorials are lacking in how they teach you blending phonemes, especially in Piapro. Piapro doesn't allow for overlapping notes like the Vocaloid 6 editor does. SynthV also doesn't allow overlapping notes, but unlike SynthV, Piapro doesn't allow you to move phonemes, meaning you often have to add the consonant sound of the next word to the end of the first, or have a "transition" consonant or vowel sound to make the slide into the next word less choppy.
In V6, you can slide the end of one note into the area of the next to let the editor know you want those sounds to transition smoother than normal, and it most cases it works wonders!!
IDK, I don't have the time to make a tutorial on it, but I might make a personal dictionary of how certain words are pronounced in my region and how I got Miku to pronounce the words.
tangentwithtanner
yousharknotpass
romaincrisis
jacobvanloon
gummygunk
baratheonpilled
jbbartram-illu
janainaart
