After some searching, I decided that the internet might like a text resource on making CV 'loids sing in english. I'm not by any means an expert, but my futzing around may have yielded results that people may find helpful. If you have any corrections, suggestions, additions, then please post below.
If you're interested in working with UTAU for english, then your best bet is CV VC. However, if you want a specific CV 'loid to sing in english, read on. This probably isn't the only method, it probably isn't even the best method: but if it sounds new, give it a whirl, eh?
a. Keyboard Shortcuts
b. Japanese Syllable Substitutions
c. Step 1: initial notes
d. Step 2: consonants
e. Step 3: Whittling it down
a. Keyboard Shortcuts
This method involves a bunch of futzing around with preutterances, lengths, overlap, so please consider the following shortcuts:
Ctrl+E â note property, where you can change the length, preutterance, overlap, and other fun things. This is an important dialog.
Ctrl+Y â envelop dialog, where you can muck about with the volume envelop and fix (!) notes.
Ctrl+G â OTO editor. When you select a note and press Ctrl+G, the syllableâs sample and OTO points will appear.
Shift+Del â this shortcut removes selected notes from the piano roll.
Take note! In the note property (Ctrl+E) the length is defined as a number. The length L4 (quarter note, default length) is defined in this dialog as "480". The smallest length UTAU will tolerate is L64, or "30".
b. Japanese Syllable Substitutions
First of all, again, if Iâve made a mistake here, please tell me. I ainât an expert.
Many CV lists are based on a Japanese list, so thatâs why this section is here.
Because itâs not always immediately obvious what syllables to enter, Iâm here to suggest which ones you choose. Ultimately, you have to listen to what your UTAU is pronouncing â thatâs what it all comes down to.
Iâll say it again: listen to what your UTAU is saying.
If youâre feeling lost then say the phrase youâre trying to create to yourself, and pay attention to what you are saying.
Anyways, firstly: English has 10 vowel sounds â long and short a e i o u. Japanese has 5 â a i u e o. Here are my suggestions for vowel substitutions:
Long vowels:
a (ay, think âhayâ) â ãï¼ã (e then i)(dipthong)
e (ee, think âkneeâ) â ã (i)
i (eye, thinkâ¦âlieâ) â ãï¼ã (a then i) (dipthong)
o (oh, think âboneâ) â ã (o)
u (oo, think âyouâ) â ã (u)
You can simulate a dipthong vowel by putting the two vowels together, extending the preutterance of the second part to 60-120 ms, and pressing the p2 p3 crossfade button.
Short vowels:
a (ah? think âbatâ) â ã (a)
e (eh, think âbetâ) â ã(e)
i (ih, think âbitâ) â ã (e)
o (ah? think âbotâ) â ã (a)
u (uh, think âbuttâ) â ã (a)
As you can see Japanese a or Japanese e will usually do the trick, but listen to the way you say it. When in doubt, trial and error is the way to go.
Japanese also doesnât have consonant sounds analogous to English â most obviously, it lacks a âthâ. There are substitutions for this and other (more egregious) consonants.
Remember that if a consonant doesnât sound right, you can try another syllable with the same consonant. So, if youâre looking for a good âlâ, then you have five syllables to choose from â ra, ri, ru, re, ro. You can pick and choose the one that your UTAU pronounces the most like âlâ.
Consonants
th â sub s- or z-.
b â use b-
v â sub b-
Note that if your UTAU is Japanese, your âloids pronounciation of ba-bi-bu-be-bo may range between b or v sounding pronunciation. When comparing your set, note that a b sound tends to have a shorter preutterance than a v sound.
f â use f- or sub b-
I find that the Japanese pronunciation of âfâ can sometimes be almost nonexistent. If you want to take it up a notch, a v-sounding âbâ might work.
s â use s- or sub z-
For plural ending english s, z- is often preferable.
r â use r- or sub w-
If you need a hard r, and your 'loid's r-s are more l-sounding, then w- is close enough.
l â sub r-
like b-, your r-s might range from âlâ to ârâ
c. Step 1: initial notes
Okay, let's get going.
When I get ready to enter English I usually put down CV bases first. Hereâs a test phrase â âthis is a frickinâ penâ. I would grind this down to âze e e we ke peâ.
You grind it down like this: You break down the phrase into syllables:
this is a fri ken pen
Take off the ending consonants:
thi i a fri ke pe
Chop off any extra consonants at the beginning:
thi i a ri cke pe
and, naturally, convert to CV consonants.
ze e e wi ke pe.
d. Step 2: consonants
Next you create notes where your 'loid pronounces the consonants you eliminated in the preliminary stage. These notes are short enough that the note is bascially all preutterance -- the consonant.
For the above sentence ("this is a frickin' pen), I would add (these) notes in at L64 (Len. 30):
ze (zu) e (zu) e (bu)wi ke (n) pe (n)
Again, Each consonant sound should be 100% preutterance. Itâs set to a low length like Len. 30 so that the vowel part doesnât have time to be pronounced.
In the case that you have two consonants next to each other:
Ex. 1 (two words): Ze (zu) sa
Ex. 2 (one word): Za (zu) (su)
The preutterance of the âsâ will be cut off by the severely cut down âzuâ, and the word won't be understandable. To fix this, you add a rest in between:
Ex. 1 (two words): Ze (su) (R) sa
Ex. 2 (one word): Za (su) (R) (su)
A rest which is as long as the note's actual preutterance. When doing this, set your Quantize to 64 and your Length to L4, for convenienceâs sake.
e. Step 3: Whittling it down
After you add your consonants and rests, you should cut down the syllables you added initially (ze e e we pe) to accomadate what you added. For example, âcry babyâ might look like this:
(ku)wa (i) (R) be bi
âbeâ and âbiâ by themselves have the right timing, but the extra consonants in âcryâ will set them off time. To fix this: Letâs say that âwaâ is Len. 480, and that (R) and (ku) and (i) are all Len. 30. All together, the added notes add up to Len. 90. 480-90 = 390, so we change âwaâ to Len. 390 in the Note Property window (Ctrl+E). Now all the words are on-time.
Give it a spin. Does it sound about right? Cool.
...And that is my entire method. I hope this helps. Please tell me if any clarification or corrections are needed. If you have a suggestion to improve this method/tutorial, please post here.
If you're interested in working with UTAU for english, then your best bet is CV VC. However, if you want a specific CV 'loid to sing in english, read on. This probably isn't the only method, it probably isn't even the best method: but if it sounds new, give it a whirl, eh?
a. Keyboard Shortcuts
b. Japanese Syllable Substitutions
c. Step 1: initial notes
d. Step 2: consonants
e. Step 3: Whittling it down
a. Keyboard Shortcuts
This method involves a bunch of futzing around with preutterances, lengths, overlap, so please consider the following shortcuts:
Ctrl+E â note property, where you can change the length, preutterance, overlap, and other fun things. This is an important dialog.
Ctrl+Y â envelop dialog, where you can muck about with the volume envelop and fix (!) notes.
Ctrl+G â OTO editor. When you select a note and press Ctrl+G, the syllableâs sample and OTO points will appear.
Shift+Del â this shortcut removes selected notes from the piano roll.
Take note! In the note property (Ctrl+E) the length is defined as a number. The length L4 (quarter note, default length) is defined in this dialog as "480". The smallest length UTAU will tolerate is L64, or "30".
b. Japanese Syllable Substitutions
First of all, again, if Iâve made a mistake here, please tell me. I ainât an expert.
Many CV lists are based on a Japanese list, so thatâs why this section is here.
Because itâs not always immediately obvious what syllables to enter, Iâm here to suggest which ones you choose. Ultimately, you have to listen to what your UTAU is pronouncing â thatâs what it all comes down to.
Iâll say it again: listen to what your UTAU is saying.
If youâre feeling lost then say the phrase youâre trying to create to yourself, and pay attention to what you are saying.
Anyways, firstly: English has 10 vowel sounds â long and short a e i o u. Japanese has 5 â a i u e o. Here are my suggestions for vowel substitutions:
Long vowels:
a (ay, think âhayâ) â ãï¼ã (e then i)(dipthong)
e (ee, think âkneeâ) â ã (i)
i (eye, thinkâ¦âlieâ) â ãï¼ã (a then i) (dipthong)
o (oh, think âboneâ) â ã (o)
u (oo, think âyouâ) â ã (u)
You can simulate a dipthong vowel by putting the two vowels together, extending the preutterance of the second part to 60-120 ms, and pressing the p2 p3 crossfade button.
Short vowels:
a (ah? think âbatâ) â ã (a)
e (eh, think âbetâ) â ã(e)
i (ih, think âbitâ) â ã (e)
o (ah? think âbotâ) â ã (a)
u (uh, think âbuttâ) â ã (a)
As you can see Japanese a or Japanese e will usually do the trick, but listen to the way you say it. When in doubt, trial and error is the way to go.
Japanese also doesnât have consonant sounds analogous to English â most obviously, it lacks a âthâ. There are substitutions for this and other (more egregious) consonants.
Remember that if a consonant doesnât sound right, you can try another syllable with the same consonant. So, if youâre looking for a good âlâ, then you have five syllables to choose from â ra, ri, ru, re, ro. You can pick and choose the one that your UTAU pronounces the most like âlâ.
Consonants
th â sub s- or z-.
b â use b-
v â sub b-
Note that if your UTAU is Japanese, your âloids pronounciation of ba-bi-bu-be-bo may range between b or v sounding pronunciation. When comparing your set, note that a b sound tends to have a shorter preutterance than a v sound.
f â use f- or sub b-
I find that the Japanese pronunciation of âfâ can sometimes be almost nonexistent. If you want to take it up a notch, a v-sounding âbâ might work.
s â use s- or sub z-
For plural ending english s, z- is often preferable.
r â use r- or sub w-
If you need a hard r, and your 'loid's r-s are more l-sounding, then w- is close enough.
l â sub r-
like b-, your r-s might range from âlâ to ârâ
c. Step 1: initial notes
Okay, let's get going.
When I get ready to enter English I usually put down CV bases first. Hereâs a test phrase â âthis is a frickinâ penâ. I would grind this down to âze e e we ke peâ.
You grind it down like this: You break down the phrase into syllables:
this is a fri ken pen
Take off the ending consonants:
thi i a fri ke pe
Chop off any extra consonants at the beginning:
thi i a ri cke pe
and, naturally, convert to CV consonants.
ze e e wi ke pe.
d. Step 2: consonants
Next you create notes where your 'loid pronounces the consonants you eliminated in the preliminary stage. These notes are short enough that the note is bascially all preutterance -- the consonant.
For the above sentence ("this is a frickin' pen), I would add (these) notes in at L64 (Len. 30):
ze (zu) e (zu) e (bu)wi ke (n) pe (n)
Again, Each consonant sound should be 100% preutterance. Itâs set to a low length like Len. 30 so that the vowel part doesnât have time to be pronounced.
In the case that you have two consonants next to each other:
Ex. 1 (two words): Ze (zu) sa
Ex. 2 (one word): Za (zu) (su)
The preutterance of the âsâ will be cut off by the severely cut down âzuâ, and the word won't be understandable. To fix this, you add a rest in between:
Ex. 1 (two words): Ze (su) (R) sa
Ex. 2 (one word): Za (su) (R) (su)
A rest which is as long as the note's actual preutterance. When doing this, set your Quantize to 64 and your Length to L4, for convenienceâs sake.
e. Step 3: Whittling it down
After you add your consonants and rests, you should cut down the syllables you added initially (ze e e we pe) to accomadate what you added. For example, âcry babyâ might look like this:
(ku)wa (i) (R) be bi
âbeâ and âbiâ by themselves have the right timing, but the extra consonants in âcryâ will set them off time. To fix this: Letâs say that âwaâ is Len. 480, and that (R) and (ku) and (i) are all Len. 30. All together, the added notes add up to Len. 90. 480-90 = 390, so we change âwaâ to Len. 390 in the Note Property window (Ctrl+E). Now all the words are on-time.
Give it a spin. Does it sound about right? Cool.
...And that is my entire method. I hope this helps. Please tell me if any clarification or corrections are needed. If you have a suggestion to improve this method/tutorial, please post here.