#utau tutorial series

I’ll only be showing how to oto CV and VCV voice banks as they are the only ones I work with.

To begin otoing we will need to go to voice bank settings which is under tools:

This will open a display which will look similar to this:

The recordings you’ve put in your utau’s folder should appear in the Name column. In my case, I’ve just got a,i,u,e,o at two different pitches. 

The next column, Alias, is used to encode your utau and is a fairly straight forward process for CV voice banks. For VCV voice banks aliasing isn’t as easy and I’ll explain my process a little later on.

To begin otoing click on the syllable you want to oto and press ‘Launch editor’.

In the top left corner we can see the name of the file and the alias in speech marks. This is helpful when when otoing VCV when you have multiple syllables in a single recording. 

The blue lines are the SOUND WAVE, you’ll be able to see this if you’re in sound editing software like audacity. 

The orange line is your TUNING. The line in the middle is where your orange line should be approximately. If you find that in your recording the orange line is hugely off the mark or dips really far at random points in the recording, it’s probably a good idea to re-record as it will noticeably affect how your utau will sound. 

The red line is the PREUTTERANCE. This is used to identify where the beginning of the vowel begins or, for VCV, where the beginning of the syllable is. 

The green line is the OVERLAP. This is the area in which will be used to blend the two sounds together.

For now just drag preutterance and overlap lines over into the open, if you continue dragging from the side you will get a pink shaded area and blue shaded area.

The pink shaded area is the CONSONANT. To prevent confusion, I will call this the CLIP. This is the part of the recording that UTAU will not stretch or as I said before, preserve the sound.  

The blue shaded area is the CUT OFF. This is the part of the recording that will be completely ignored by UTAU.

Once you’ve dragged out everything, It should have all these parts.

CV Otoing process

It’s best to start with the CUT OFF as all your values will move with once they’ve been moved out. This should be placed where the sound wave begins, I find it’s nice to just give it a tiny little bit of silence at the beginning.  If you go to the end of the recording you can also drag a cut off area to cut off the end of your recording. If you had the problem where you go off key at the end of your recording, this is where you can fix it. Just cut off the recording just before you orange line starts bending weirdly. But make sure that it’s not too close to the Clip shaded area otherwise UTAU will have noting to stretch! 

Next is the CLIP (pink shaded area), this should cover the consonant and vowel as this is the bit you want to be able to hear clearly. Drag it all the way over till there is a semi consistent thickness in the sound wave in the vowel. 

The RED LINE/PREUTTERNACE should be where the beginning of your vowel is. With vowel recordings like a, i, u, e and o, you’ll want to place it right at the beginning of the sound wave. With syllables like ‘ka’, you’ll want to drag it to where the vowel begins. To identify this, usually there’ll be a bit of a bump in the sound wave where the consonant and usually where then next raise is where the Vowel is. 

 The GREEN LINE/OVERLAP should be placed about half way between the CUT OFF and the PREUTTERANCE.  

Once you’re done it should look something like this:

A good way to test if your oto is working is to get your Utau to sing in progressively shorter intervals for each line of the hirigana alphabet. e.g. ka, ki, ku, ke, ko. e.g.2. ta, chi, tsu, te, to. etc...

Example:

If you’ve correctly oto’ed you should be able to clearly hear each syllable at each interval without it sounding like the end has been chopped off.

CV Aliasing

Aliases are used to encode your utau in hirigana and romaji.

For example: ka.wav = か

At the moment, there is no way to automatically alias all your recordings and it can only be done manually. This can be done quicker if you use the oto.ini file in your utau’s folder.

In this file you should see something along the lines of:

a.wav=,,,,,                          From this you should change it to the following: e.wav=,,,,,                          a.wav=あ,,,,, i.wav,,,,,                             i.wav=い,,,,, o.wav,,,,　　　　　　　　o.wav=お,,,,, u.wav,,,,,                            etc...

If you’ve already oto’d your recordings before aliasing, just put the hirigana character before the commas. e.g. a.wav=あ, 37, 1254, 90, 564, 0.

If your computer won’t let you save in your utau folder (it should but if not) save it somewhere else, but Make sure to save your oto as a .txt file in UTF-8 encoding or else the hirigana will end up as strange symbols. You can change the .txt encoding here when saving:

Once you’ve done that with the syllables, open your new oto in UTAU by going to files.

Find the oto.ini file and your alias should appear like so:

If you had to save your oto elsewhere, make sure you’re searching with files with .txt and not .ini.

VCV Otoing process

With VCV or Vowel-Consonant-Vowel voice banks we want to use the otoing to chop up our recording to get the clips we need. Because there are multiple clips to one recording we’ll need to duplicate it in the oto table. Unlike CV aliasing, UTAU will not recognise the duplicates if you do it directly in the oto.ini file.  Do NOT make duplicates of your recording (.wav) in your utau’s folder. 

To do duplicate, select the .wav file you want and press duplicate.

It is possible to alias in both romaji and hiragana, however it would double your work load so until you’ve aliased fully in one or the other. I wouldn’t recommend aliasing in both.

If you’ve never oto’ed VCV before it can be a bit confusing to understand. But, what we do is use the Vowel from the previous syllable and stick it onto the current syllable. Thus why there’s in the alias it’s [o ヴぉ]. 

I like to duplicate the non-oto’ed version of the recording and Alias it before I actually start the oto. I find if I put each syllable in first and then add the vowel of the previous one it’s a little faster. Make sure to press set every time you put an alias.

When you open up your editor it should look along the lines of this (If you zoom out):

If you don’t know how to read sound waves, we can see how each part breaks up.

So when we oto, we oto it like this:

let’s work with [u va] or [u ヴぁ]! (the second section of this recording)

For the CUT OFF we want to bring it to the beginning “u” sound from the previous “Vu” sound. This is usually around the middle of the “Vu” sound, Or, much like finding the Vowel for CV, where the shape of the Sound wave changes. You can shift the cut off area around and double click on the ‘P’ button in the top left corner next to the + and - buttons, this will playback the current oto of your recording (excluding the Cut off areas).

With the END CUT OFF, you want to Bring it to the end of the “Va” syllable.

With the RED LINE/PREUTTERANCE you want place it at the beginning of your syllable. This will make sure that the syllable “Va” is sung in time rather that being playing “u va” and the “va” sound being heard at the wrong time.

With the GREEN LINE/OVERLAP, again, you want to place it in between the Preutterance and Cut off, however, more often than not you will want to place it closer to the cut off than you would with a CV as it will give more Vowel for UTAU to blend with. 

With the CLIP/PINK SHADED AREA you want to have it cover the previous Vowel and the entire syllable. This will make sure UTAU will use “u va” instead of  “u v-”. As you can see from the example above, this does not mean you need to cover the entire section, where the thickness of the sound wave is consistent is approximately where the vowel is extended. 

Continue this process with the rest of the recording. With the beginning segment of the recording, oto it like you would with a CV recording making sure to cut off the following syllable. An easy way to just go down the list is to use the arrows at the bottom of the editor, this will go up and down the list for you to oto each syllable with out having to exit the editor. 

To check if you’ve oto’ed correctly, you can line up the syllables you’ve just oto’ed. e.g. - vu, u va, a va, a vi etc...

Here I’ve got the recording’s syllables repeated 3 times.

1) 1/4 notes to see if the sound sounds smooth. Make sure you cross fade

2) alternating notes progressively getting higher to test smoothness between notes.

3) 1/8 notes to see if you can clearly hear each syllable. 

Don’t be afraid of the envelope going over the end of the note’s area. This is how the syllables blend together, if the Vowel is cut off for a syllable it will be replaced by the overlap for the one following after. 

If you find that when you add new files to your utau folder while you oto other files, you may find that the new files don’t show up. Don’t panic! Simply refresh by going to File Refresh.

If you have any questions feel free to ask me!