How To Make an UTAU

Create an UTAU

Before you begin reading it's very important that you know the rules about making an UTAUloid. I'll list them below, and I found them in this guide to make your own UTAUloid which you can download here : http://u9.getuploader.com/utaou/download/72/How+To+Create+your+own+UTAU+voice+bank+rev0.40.pdf

"You may enjoy yourself, but do not break the law or violate somebody else's rights.
All "UTAU" users must follow these rules!!
•Don't create voice bank from a real singer's voice without permission.
•Don't create voice bank from a real actor/actress's voice without permission.
•Don't create voice bank from a real voice actor/actress's voice without permission.
•Don't create voice bank from the output of “Vocaloid” products, which explicitly forbid such a usage.
•Don't create voice bank from the output of other voice synthesizers without permission.
Breaking the rules will result in the accusation against you, and may even Mr. 飴屋／菖蒲 (Ameya/Ayame) as an accomplice.
Such a situation will terminate the free "UTAU" world and should be avoided definitely. "

I'm assuming you all have the English patch for UTAU, if you don't go to the downloads section and download it. Also there's a new version for UTAU 2.61 http://ux.getuploader.com/utau_a5/download/33/utau_a5_33.zip.
Here are some reclists http://bit.ly/HiraRomaListsCV_VCV
full short and CV lists Hiragana>Romaji courtesy of YoruIkusane
Well in UTAU go to Tools > Voicebank Settings (or ctrl G) and you get this window. Configure everything by following the explanation below and using your own judgement and then click OK and the oto.ini file is AUTOMATICALLY created in your selected voicebank.

Alias is the alternative name you will give your samples. If you saved your wav files as romaji, type the hiragana as the alias. If you saved your wav files as hiragana, type the romaji as the alias.

Offset is the parameter of the blue highlight thing from the left. Basically you can cover any blank space on your wav file or any weird sound from your recording with this and it wont be heard when you make your UTAU sing.

Consonant is the parameter of the pink highlight thing. Anything you cover with this will not be altered when the note is stretched in UTAU.

Cuttoff is the parameter of the blue highlight from the right. Just cover any blank space after your recording with this or anything that is not consistent with the rest of your recording at the end.

Preutter is the parameter of the red line. Put this at the end of the consonant in your recording. When you select the red line in the editor you will notice it says "top of note" this is basically how soon the note will begin to sound when you make your UTAUloid sing. This parameter shortens the lenght of the preceding note.

Overlap is the parameter of the green line. This parameter defines the length of utterance extention of the *preceding* note in milliseconds. Sometimes this parameter isn't needed at all but when the consonant is too long I usually put this at the middle of the consonant in the recording. With this you are assigning how soon the note with this sample will overlap with the previous note.

On that window, select the sample you are going to configure then click on "Launch editor". You should see a window with your wav file. Then you can move around some parameters that will write in the numbers of the oto.ini
In summary the blue covers all that you don't want to be heard, or you can use it to cut off empty space. The pink covers the area that you don't want to be manipulated when you stretch a note so cover the consonant and part of the vowel. The green line determines when the note will overlap with the previous note (it shortens the previous note too) and the red line says how much it will get ahead. Just place the green bar in the middle of the consonant and the red bar at the end of the consonant.

Also the highlighter like things are hidden drag the red and green lines first then you'll be able to drag the pink thing and after you drag the pink you can drag the blue. You can also drag the pink and blue before dragging the green and red lines by clicking on the leftmost corner where there is no green or red line visible. Also if you drag from the right, you immediately get the blue highlight thing.

Oh something else I was told (I believe it was Nami who figured this out) On the editor window (where you drag everything around) there is a button with a "P" on it. If you click on that it'll let you listen to the sound once configured.

How To Make Oto (Basic Tutorial)

Well in UTAU go to Tools > Voicebank Settings (or ctrl G) and you get this window. Configure everything by following the explanation below and using your own judgement and then click OK and the oto.ini file is AUTOMATICALLY created in your selected voicebank.
http://i50.tinypic.com/2rh751j.png (click to view image)

Alias is the alternative name you will give your samples. If you saved your wav files as romaji, type the hiragana as the alias. If you saved your wav files as hiragana, type the romaji as the alias.

Offset is the parameter of the blue highlight thing from the left. Basically you can cover any blank space on your wav file or any weird sound from your recording with this and it wont be heard when you make your UTAU sing.

Consonant is the parameter of the pink highlight thing. Anything you cover with this will not be altered when the note is stretched in UTAU.

Cuttoff is the parameter of the blue highlight from the right. Just cover any blank space after your recording with this or anything that is not consistent with the rest of your recording at the end.

Preutter is the parameter of the red line. Put this at the end of the consonant in your recording. When you select the red line in the editor you will notice it says "top of note" this is basically how soon the note will begin to sound when you make your UTAUloid sing. This parameter shortens the lenght of the preceding note.

Overlap is the parameter of the green line. This parameter defines the length of utterance extention of the *preceding* note in milliseconds. Sometimes this parameter isn't needed at all but when the consonant is too long I usually put this at the middle of the consonant in the recording. With this you are assigning how soon the note with this sample will overlap with the previous note.

On that window, select the sample you are going to configure then click on "Launch editor". You should see a window with your wav file. Then you can move around some parameters that will write in the numbers of the oto.ini. Basically it should look something like this.
http://i46.tinypic.com/307nuk7.png (click to view image)
In summary the blue covers all that you don't want to be heard, or you can use it to cut off empty space. The pink covers the area that you don't want to be manipulated when you stretch a note so cover the consonant and part of the vowel. The green line determines when the note will overlap with the previous note (it shortens the previous note too) and the red line says how much it will get ahead. Just place the green bar in the middle of the consonant and the red bar at the end of the consonant.

Also the highlighter like things are hidden drag the red and green lines first then you'll be able to drag the pink thing and after you drag the pink you can drag the blue. You can also drag the pink and blue before dragging the green and red lines by clicking on the leftmost corner where there is no green or red line visible. Also if you drag from the right, you immediately get the blue highlight thing.

Oh something else I was told (I believe it was Nami who figured this out) On the editor window (where you drag everything around) there is a button with a "P" on it. If you click on that it'll let you listen to the sound once configured.
Configuring the oto.ini for a VCV voicebank

First you need to get recording so here is a short list by YoruIkusane: Short VCV list

It took me a while to finally understand how to configure the oto.ini of a VCV voicebank but thanks to mianaito's help I finally understand how.

Here is what he basically told me "Essence of VCV configuration is: 1)overlap=end of 1st Vowel 2)preutter=end of Consonant. Consonant lenght is not important."
So basically you can start the configuration by setting the consonat lenght and the Cuttoff with fixed values then fix the rest as needed so that it looks something like this:
http://i46.tinypic.com/2yjv3uh.png (click to view image)
When mianaito helped camila configure Camila's VCV he started with the consonant as 420 and the cuttoff as -700. You can use SetParam (which comes with OREMO) to edit those in bulk.

I hope you guys understand now how to use an oto.

Make The Voice Bank 

When it comes to recording I always suggest making each sample 2 seconds long and using the program you are using to record (Audacity for me) to fade out a little bit of the end.
If you are looking for a relist (a list of things to record) I just searched hiragana on Wikipedia and recorded according to that.
You want to save all of your samples in the same folder, just remember that.

When importing your samples in to UTAU, click T then O then the second tab in the pop-up menu.
You will see a large box then two smaller input boxes below it.
The bottom-most input box has a button to it's right, click it.
Find the folder that your UTAU samples are in then click on one of your samples then open.
In the input box above the one you just used, is where you put the name of your UTAUloid then click the left most button on the right of the box.
The right most button is the delete button.
Click OK when your are done with this.

Now you have to make the .frq files.
Click T then S (or ctl + G) and a pop-up menu will open with all of your samples listed.
You will have to click sample individually then click the second button to the right.
Click okay on the pop-up then wait.
When you are done waiting, a little circle will appear on the right most part of the large box under frq.
You will have to do this for each sample you recorded.
Once all of the samples have circles, click okay at the bottom of the box.

You can start using your UTAUloid after that.

Recording Your ACT2 Voice Bank 

This is for recording an ACT2 VB (VB thats re-recorded for better results), but some voice recording tips are also written in here for beginner UTAUs (i didnt recommend fresh-started ppls to use this since u still dunno which sample sounds good in UTAU) oh one more thing, if you haven't learn about setting the oto.ini, i suggest you read it 1st before this (look on the photos, i had it stored there) 

Alright, have you made your VBs peoples? Now atleast you've used your VB to make couple of songs, so i assume you already know what samples that sounds weird on the program (sometimes even good samples sounds weird in UTAU). So its time for ACT2! Well anyways i can't teach u much about this matter yet cuz i'm still having my own ACT2 in progress xD Some peoples also add english VB to ACT2, but some UTAUs sounds like asdjhlfjdkalkj when they sing in eng.

Okaay, now i assume u had an idea 'how' your samples sounds when its synthesized right?? most of peoples had problems with the samples that ends with 'i', mine sounds a bit robotic usually good ust can cover these kind of minor error, but not every ust is good. 

back to audacity. Note down the samples thats broken/sounds weird. Make sure thats THE SAMPLE is weird, NOT the oto.ini settings. after you noted all of 'em, clear the oto.ini's for each samples and delete each .frq (current sample config, if u dont delete it the new sample will sounds hellish) and .frt (i have no idea what issit but just delete it) files of your samples. Open the .wav files and try to listen to it (on audacity). Perhaps you know whats the problem now. delete it naw and start recording.

NOTE. Defoko/Uta Utane(the defalut VB)'s samples didn't came from human voice and it sounds sorta airy and robotic. DON'T use it as your recording example, I reccomend using professional, popular, clear voiced and well tuned UTAUs like Camila Melodia, Namine Ritsu, Utaune Nami, Hoshine Stella, etc. oh if you wanna use the popular VB's as example, make sure you download the right type. DON'T use a VCV VB for an example to record CV VB. and don't use another UTAU Voice Bank! (sample you don't use another UTAU Voicebank like Utaune Nami)

NOTE 2ND. After you made an ACT2, don't expect it will sound as good as vocaloids. it will still sound a bit heavy like the other UTAUs. Chill, you're not the only one who had VB like that.

Samples that usually sounds good on UTAU are:
1. Clear recorded samples (example. Hanane Mizu's 'a.wav')
2. Samples with a 'bit' accents on it (too much accents will ruin it). Most of the *ya *yu *yo (example kya, byo, ryu) belongs to this category xD I suggest looking at Utaune Nami's VB for this category.
3. Well tuned & placed (sort of) robotic VB, or in other words, VBs that are made 'robotic' in purpose. Vocaloid Gumi (Megpoid) belongs here. For robotic UTAUs, i said its, Kasane Teto and Kagene Himeka. but idk how she sounds a bit robotic but 'clear', not sorta deep and heavy. sing on c5 and bye robotic voice~

Okay, had some troubles recording certain samples? Here are some tips
1. for example you had trouble recording 'ni'. try to listen to your 'na' 'nu' 'ne' 'no' (the one that sounds the best) and also try to listen a word with vowel 'i' behind. Try to imitate it!
2. for the ya yu yo samples (like kya byu gyo, etc), you can pronounce kiya with the ki shortened instead. If its hard, normally pronounce kiya and cut off the 'i'
3. You can try to use an english word/japanese word to pronounce the syllabels you want as the 'word' and cut it from the rest. For example to record my 'ryo' i said matoryoshka and cut off the ryo seriously  why i said matoryoshka instead matryoshka? its easier to cut the ryo from to-ryo than tryo. in my opinion, the samples that starts with 'r' are pretty hard to pronounce (cuz japanese style of r sounds sorta like r n l merged, it sounds like 'd' instead), so i used japanese word to record some of them. for example u can use 'dakara' for ra, kirei (read, kiree) for the re. mmm in english, for example you can say 'one' and cut off the 'n' behind. be careful in cutting it tho!

Now, try it out in UTAU! (it'll need some time cuz it will re-create new frq files)
if it sounds good, congrats! set up the oto.ini and you're done!

if it still sounds horrible, delete the frqs and re-record again. don't worry if it doesnt sound right.

That all the ways to How to Make an UTAU. Hope you understand. i'm not have an UTAU. i got it from my friends :) make an UTAU must have a microphone. but i don't have that :'( that because i'm really want to make an UTAU! ANYONE, HELP ME!!!