Essential Tuning Skills

Arissa

Ritsu's Renegades
Defender of Defoko
Hello, everyone, been a while since my last discussion post. Since I'm still somewhat of a noob at tuning, I'd like to discuss it. I want to know

•What do u believe is essential to make an Utau sound good
•To make it sound human
What things do you use in the program, like how do you use utilize curves, plugins, flags, all your go-to's, resamplers, breaths, cross-fades, and for which types of banks? Like soft, strong, nasally, I want to know everything about what you use when tuning and why >=<

I use A la carte at times, but I'm no pro,, I also use cross-fades(mainly been working with CV) but am not a pro,,

Depending on what's up I reset a ust completely to give it a full retune, although I haven't lately...
 
  • Like
Reactions: RenzoA

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
I'm rather lazy at expressive tuning, so this is more like pre-tuning advice:
  • Multipitch voicebanks are excellent
  • In CVVC banks, take full advantage of transitional notes
  • Mind phoneme choice, how words and sounds blend into each other
    • It's more important to have something sound natural and expressive than for there to be hyper-clear pronunciation! For example "would'ja" sounds better than "would you"
  • Pay attention to the quirks of the specific voicebank (even adjusting oto if necessary)
  • Make sure everything has a pitchbend, and maybe put a weak vibrato on all notes longer than a quarter. My settings:upload_2017-6-11_19-57-3.png (You could even add more control points in the Portamento section to make it easier to tune, so that you don't have to create new points all the time)
More on the mixing side, I'd say that standalone breaths should be integrated after rendering. When you put them through UTAU by itself, you get distortion artefacts from the resampler (unless the resampler you're using has a direct pass-through option). And it's quite hard to time them realistically in a UST! So, stick to adding breaths while mixing. As for end breaths, they're designed to work within the voicebank, so those should be fine to integrate.

I'm kinda lazy with resamplers, but at most I might use Moresampler, because of the Mt flag (which can simulate power/soft!) Also, moresampler's method of synthesis is better at handling banks with short samples, so I may also switch to it if the bank I'm using is lower quality and causes a lot of stretching artefacts in other resamplers. Like in this cover: https://clyp.it/ygvd5d3n I used flags to exaggerate the softness and power of the two voicebanks I was using, and it wasn't like I could ask the creator to make entirely new voicebanks just for me LOL
 

Arissa

Ritsu's Renegades
Defender of Defoko
Thread starter
Imo, even though this isn't to do with the utau application, I think mixing is one of the most important parts of making the utau sound good. Possible the most important part for me
Care to say why? And dang I was gonna include to not talk about mixing xD
 

SOHBlue

Master of Doom
Defender of Defoko
I would personally say that playing with vocal consistency is important, as in the tone and pitch. Human voices can't be perfect and it's the warmth and bit of randomness that makes it sound human. I'm not saying go nuts with pitchbends (although that can be fun and help) but say/sing things to yourself.

Think about how your voice softens at the end of a sentence or how you start off singing a note a bit sharply before evening out. Especially in speech our voices will get higher and lower depending on what we're feeling and as long as you're subtle but purposeful with the changes it can sound convincing.
Even if you think a word or lyric just sounds level, make one higher or lower, have the pitchbend sink and rise between them, add a gentle softening vibrato at the end of it. Just don't let it be too perfect, ya know?

Besides that, mixing and crossfading are very very important, I agree. Also keep in mind that often words are shortened. Like though "family" is pronounced "fam-ill-lee", many people just say "fam-ly" and to me that also sounds more natural and human depending on your tone and way of speaking.

In short, keep the pronunciation and vowels flowing and understandable, but don't be afraid to play with how your words are delivered. (Also keep in mind that I primarily work with hanasu-though not the plugin, I mean UTAU talkoid-since I prefer singing myself. But I think the same thing applies. I also love playing with modulation and intensity as well as breathiness, but that's me. Flags are also super important to me, I never tune without them. The one I got from Kowareta was : C99c99Y9
I use it all the time and it really helps as a good base as it makes the voice more clear, but I also use breathiness like I said and moresampler's Mt for tenseness and sometimes the one for how open the mouth is. *looks it up* Mo for openness.

Hope these tips make sense and help at least somewhat!
[doublepost=1497226244][/doublepost]
Care to say why? And dang I was gonna include to not talk about mixing xD
I honestly want to learn more about mixing, I was thinking about starting a thread for tips myself...
 

taigz

dead
Defender of Defoko
And y'all guys said everything.

Welp, I'm not a pro either, but I'd discuss stuff anyway.
So, here's the things I use for UTAU to sound, well not human, but at least better sounding:

Pitchbends -- The complex part in tuning, but the most powerful. In pitchbending, instead of using the pitch editor plugin, I edit directly on the piano roll because... I just do. Whatever floats your boat, but it's just what I do. Btw, the zoom in option is helpful for short notes. But for really short notes that are difficult to add pitchbends with, I just leave them be. As for the curves, I mostly use S, rarely use R and J, and I never use Linear.

tuned_ust.png

(my tuning still sucks but uhhh)
untuned_ust.png

And here is what they sound like (untuned comes in first): https://clyp.it/tymbyqvx

Which sounds more emotional? (obivous answer is obvious)

To be honest, pitchbending is hard at first... but well it's actually not. And it's pretty fun! (i'm just too lazy to finish my tuning wips lel)

Flags and Resamplers -- They are also super important. For resamplers, I use Moresampler, the default, Fresamp, vs4u, TIPS and that's all. Moresampler tho is really good (but not for those who sound like satan with it, lel) as it has those good extension flags (Mp is favorite nah jk it's Mt). Tbh here's the thing: if you see flags, try it out on some UTAU, or experiment with them yourself. Same with resamplers... oh glob I don't have much to say about this. *cries*

I was going to include mixing but I'll save that for later...
 

WyndReed

Dareka tasukete!
Global Mod
Defender of Defoko
Well, to start with, there is no "one way to make utau sound good". Everyone has their own preferences as to what sounds right to them, and tuning an ust really is about getting the sound right for you.

This is my take on/what I do with tuning:

  • For the most part I use utau banks with out flags except for some spot treatment here and there.(maybe play with gen and bre a little bit in certain parts)
  • I've had a hard time with some powerscale banks. I've found that some of them have more of a "I'm really angry/shrill type power" instead of an "I feel strongly about something/more calm type" of power. Which I've found is harder to use outside of Deco27's latest songs.
  • Multi-pitch banks are great, but mono-pitch banks shouldn't be forgotten about. There are some excellent ones out there that with the right pitchbends and envelope editing can be pretty expressive.
  • There are so many resamplers out there and they all want to play differently with different utau and different types of pitch bends. Its kind of a trial and error process to figure out which ones work with which banks and also to figure out which sounds the best to you for those banks.
  • Pitch bends and vibrato are really dependent on what musical genre you're working with. For instance with a rock song, you might be more likely to use more dramatic pitchbends then in a softer fluffy pop song.
  • End breaths are good. I always have to edit the envelopes and/or volume so that I get just that "end of the sound" and not the "hh" that follows.
  • A la carte is great. I just started experimenting with it myself and have found that you can get a good base from it as long as you click the right parameters.
  • The multi-envelope editor and multi pitch editor plugins are very helpful. They seem to only want to work with certain banks/resamplers though.
  • Look at other people's ust's and how they tune. For example, you might find that someone does some creative use of vibrato that you can then pick up and add to your skill set.
  • If you are in the piano roll and press "N" on your keyboard the plugin list will pop up. (this works in win-utau not sure about utau-synth)

I guess my advice is really just to experiment until you find a style you like. There is no definite way to tune, only the way that sounds good to you.
 

Arissa

Ritsu's Renegades
Defender of Defoko
Thread starter
Well, to start with, there is no "one way to make utau sound good". Everyone has their own preferences as to what sounds right to them, and tuning an ust really is about getting the sound right for you.

This is my take on/what I do with tuning:

  • For the most part I use utau banks with out flags except for some spot treatment here and there.(maybe play with gen and bre a little bit in certain parts)
  • I've had a hard time with some powerscale banks. I've found that some of them have more of a "I'm really angry/shrill type power" instead of an "I feel strongly about something/more calm type" of power. Which I've found is harder to use outside of Deco27's latest songs.
  • Multi-pitch banks are great, but mono-pitch banks shouldn't be forgotten about. There are some excellent ones out there that with the right pitchbends and envelope editing can be pretty expressive.
  • There are so many resamplers out there and they all want to play differently with different utau and different types of pitch bends. Its kind of a trial and error process to figure out which ones work with which banks and also to figure out which sounds the best to you for those banks.
  • Pitch bends and vibrato are really dependent on what musical genre you're working with. For instance with a rock song, you might be more likely to use more dramatic pitchbends then in a softer fluffy pop song.
  • End breaths are good. I always have to edit the envelopes and/or volume so that I get just that "end of the sound" and not the "hh" that follows.
  • A la carte is great. I just started experimenting with it myself and have found that you can get a good base from it as long as you click the right parameters.
  • The multi-envelope editor and multi pitch editor plugins are very helpful. They seem to only want to work with certain banks/resamplers though.
  • Look at other people's ust's and how they tune. For example, you might find that someone does some creative use of vibrato that you can then pick up and add to your skill set.
  • If you are in the piano roll and press "N" on your keyboard the plugin list will pop up. (this works in win-utau not sure about utau-synth)

I guess my advice is really just to experiment until you find a style you like. There is no definite way to tune, only the way that sounds good to you.
Ik that, I never said there was one way, this thread is for ppl to talk about ways they tune and make an Utau sound good, I didn't give a definite right or wrong way of tuning, I'm allowing people to express how they tune; discussing it.
 

kimchi-tan

Your local Mikotard
Global Mod
Defender of Defoko
Since a lot has been said already, I'll keep mine brief as not to repeat much what others have already said.



A lot of factors contribute into what can affect the utau's output. Here are my two cents:


The Voicebank
The samples dictate how the utau will sound and this includes the "emotion" of how it will sound. A voicebank recorded with a monotonous expression will give you a bored-sounding utau. Listen to the samples of utau that you find emotional and observe how they are recorded.

Pitchbending
Arguebly the biggest factor when it comes to tuning. One of the best ways to develop your pitchbending is to observe the pitchbending on well-tuned USTs. Just play them out and watch the playback cursor hover over each note. That way, you will slowly understand how each kind of pitchbend affects the output of the sound.

Modulation

Oh boy the part I'm most crazy about in tuning - modulation. What does it do? It pitch-corrects the samples before the resamples changes the pitch since some syllables in the samples may sound off-pitch. Mod0 is 100% pitch-corrected an mod100 is 0% pitch corrected.

Here's a sample to go along with the screenshot I posted below to show how mod affects the notes: https://clyp.it/1wkgjhvn

It's not too obvious, but the first render (which is at mod0) sounds a bit stiffer than the second one (which has the mod values of the screenshot I posted).

How much modulation you will need depends on how intensely the voicebank was recorded. The voicebank I used was recorded at a stable and neutral tone, so using higher mod values doesn't make it sound off-pitch.

upload_2017-6-12_14-28-50.png

Flags
Each resampler has a different set of flags than the others so you might want to take a look at their readme since that's usually where the resampler's flags are listed down in. Moresampler by far has the most diverse amount of flags which really change the tone of the voice.
The voicebank I used for the sample was just a monopitch voicebank so it only had 1 tone but I wanted a more dynamic result so I used moresampler's flags to my advantage. As you see in the screenshot above, I added flags to the first few notes to make them sound weaker and breathier than the succeeding notes (the default flags had Mt20Mo10Mb-40 btw) making the transition from weaker to stronger (it went from Mt-50 to Mt-10 making the transition not too drastic).

Consonants/Consonant Velocity
What consonant velocity does is to compress or to stretch the pink/consonant area of the oto since it is usually left unchanged. 0 stretches it by twice the length and 200 compresses it by half the length. It's not shown in the screnshot above, but I made the convel for the first note 0 to make it drag on longer to emphasize the "t" sound.
If you use CVVC, you won't really need to use convel much (unless it's for the first note) and you have a more visual control over the length of the consonants.
Here's another sample: https://clyp.it/xd4ztkys
upload_2017-6-12_14-58-41.png
upload_2017-6-12_14-57-31.png

For the first render, the [a sh] takes up less of the note than the second one. The second render having a longer VC gives more emphasis on the note.
This can be done with VCV voicebanks too through convel by lowering the value but sometimes you'd have to edit the oto yourself to manually create a VC depending on how much vowel takes up the pink part before the consonant on faster tempoed songs.

I'm not too versed in CVVC. How do you do this, and does it work with all languages?
Quoting this since the last one answers your question hopefully. it's a mess
 

HulderBulder

Retired User
Retired User
Defender of Defoko
Since alot has been said about how to do it, I'll name a few must have plugins:
"コピー・ペーストプラグイン" - Copy the selected part to clipboard and allows you to paste it in another ust.
"音程コピペプラグイン" - Copy and paste pitch editing.
"拡張エンベロープエディタ" - Envelope editor. Lets you edit the envelopes of all the selected notes in one go (quicker and easier since you dont have to right click every note).
"拡張フラグエディタ" - Flag editor. Lets you edit flags of the selected notes in one go (pretty much the same as envelope editor but with flags).
"拡張ピッチエディタ" - Pitch editor. Lets you draw and move pitchbends in one go (also simmilar to the envelope editor but for pitches).
iroiro - Big convertion tool. No more worries with CV and VCV compadabillety in ust.
"autocvvc 1.23(&A)" - Converts ust to CVVC, CV and VCV (I only use it for the CVVC conversion since iroiro lets you check if theres something wrong and allows you to edit then and there).
Humanizer - randomly adds a fixed ammount of modulation, intensity, and starting points of notes (I recomend not adjusting starting point since it can make the ust off time).

There are more but these are the ones I tend to use often.
you can find a list of plugins here link. its written in japanese but google can give a rough translation of the page.
 

Arissa

Ritsu's Renegades
Defender of Defoko
Thread starter
Since a lot has been said already, I'll keep mine brief as not to repeat much what others have already said.



A lot of factors contribute into what can affect the utau's output. Here are my two cents:


The Voicebank
The samples dictate how the utau will sound and this includes the "emotion" of how it will sound. A voicebank recorded with a monotonous expression will give you a bored-sounding utau. Listen to the samples of utau that you find emotional and observe how they are recorded.

Pitchbending
Arguebly the biggest factor when it comes to tuning. One of the best ways to develop your pitchbending is to observe the pitchbending on well-tuned USTs. Just play them out and watch the playback cursor hover over each note. That way, you will slowly understand how each kind of pitchbend affects the output of the sound.

Modulation

Oh boy the part I'm most crazy about in tuning - modulation. What does it do? It pitch-corrects the samples before the resamples changes the pitch since some syllables in the samples may sound off-pitch. Mod0 is 100% pitch-corrected an mod100 is 0% pitch corrected.

Here's a sample to go along with the screenshot I posted below to show how mod affects the notes: https://clyp.it/1wkgjhvn

It's not too obvious, but the first render (which is at mod0) sounds a bit stiffer than the second one (which has the mod values of the screenshot I posted).

How much modulation you will need depends on how intensely the voicebank was recorded. The voicebank I used was recorded at a stable and neutral tone, so using higher mod values doesn't make it sound off-pitch.

View attachment 5619

Flags
Each resampler has a different set of flags than the others so you might want to take a look at their readme since that's usually where the resampler's flags are listed down in. Moresampler by far has the most diverse amount of flags which really change the tone of the voice.
The voicebank I used for the sample was just a monopitch voicebank so it only had 1 tone but I wanted a more dynamic result so I used moresampler's flags to my advantage. As you see in the screenshot above, I added flags to the first few notes to make them sound weaker and breathier than the succeeding notes (the default flags had Mt20Mo10Mb-40 btw) making the transition from weaker to stronger (it went from Mt-50 to Mt-10 making the transition not too drastic).

Consonants/Consonant Velocity
What consonant velocity does is to compress or to stretch the pink/consonant area of the oto since it is usually left unchanged. 0 stretches it by twice the length and 200 compresses it by half the length. It's not shown in the screnshot above, but I made the convel for the first note 0 to make it drag on longer to emphasize the "t" sound.
If you use CVVC, you won't really need to use convel much (unless it's for the first note) and you have a more visual control over the length of the consonants.
Here's another sample: https://clyp.it/xd4ztkys
View attachment 5623
View attachment 5621

For the first render, the [a sh] takes up less of the note than the second one. The second render having a longer VC gives more emphasis on the note.
This can be done with VCV voicebanks too through convel by lowering the value but sometimes you'd have to edit the oto yourself to manually create a VC depending on how much vowel takes up the pink part before the consonant on faster tempoed songs.


Quoting this since the last one answers your question hopefully. it's a mess


Got-dang!! So much tips and advice.... Yeah, I was always curious about the modulation thingy, looks like that tool will HP with some things...(one sample I moved over sounds strange and out of tune whenever I play it;never sounds like the note it's playing) And ahhhhh just so much good advice!!!! The holy tuning master kimchi has epwo1&$! *dies
 

WyndReed

Dareka tasukete!
Global Mod
Defender of Defoko
Ik that, I never said there was one way, this thread is for ppl to talk about ways they tune and make an Utau sound good, I didn't give a definite right or wrong way of tuning, I'm allowing people to express how they tune; discussing it.
I apologize if what I said came across as preachy or something. I was mainly just trying to get my views on tuning as a whole across.
 
  • Like
Reactions: Arissa

Arissa

Ritsu's Renegades
Defender of Defoko
Thread starter

I used this as a guide to help me start off. Eventually, you get a hang of experimenting with different bends. Also if you set the modulation to 20 it helps the voice to sound a little less harsh.
Moresampler is a good with many types of banks in my experience, although TIPS is pretty good if the samples are of a lower quality or if its a deep vb.

Omg same like I watch all of Yuunari's tutorials :3 And I see, although I don't believe TIPs works with me for some reason, same with More samplersampler

And I have begun using Modulation a bit, trying out the different effects of it and whatnot, same with A La Carte
 

Zoku

making doper vocaloid music than the rest
Defender of Defoko
I think that MOD is best used in moderation.. A lot of VBs (at least the ones I've used) don't have a consistent pitch in all the samples, so I usually wouldn't go over 10% or 15%.
 
  • Like
Reactions: Arissa

☆彡иα∂ιѕ☆彡

Ash Ketchum
Supporter
Defender of Defoko
Okay, so let's talk about tuning homies

Like Kyo said, multipitch vb's are nutworthy and very helpful in tuning. And to make a bank sound smoother you can mess with Velocity and also preutterance.

As for like actual pitch bends, stay away from a la carte unless you absolutely know what you're doing because it makes the utau sound drunk and it gets rid of any nice pitchbends you have done beforehand. So either use it before tuning or not at all. But setting your own pitchbends, it really helps to listen to human versions of vocaloid covers because then you can make pitchbends that echo the utaite cover or sound similar.

You don't need any extravagant bullshit or anything, seriously, as long as it sounds natural or nice, then??? who cares how your tuning looks.

It's important to mess with pitchbends and hear what it spunds like if you pull one way down or way up because those can impact your cover greatly.

Here's a lil thing i tuned:

and this one
https://clyp.it/hrt31jvj

both using that same technique.

And if there are little covers using actual Utaites (like Rera....) then you have to be really familiar with the song and experiment to see what sounds goon.

Add more control points to your pitchbends, take some away, put some up higher, put some down lower, stretch them out, make them narrower.

Modulation is also a goodie. People are always like "Omg don't use it unless your kyaami" but tbh it's really not hard to use. I like to use 30-35 modulation and then lower it on the notes that sound gross and off tune after i'm done tuning!!
 
Last edited:

Arissa

Ritsu's Renegades
Defender of Defoko
Thread starter
Okay, so let's talk about tuning homies

Like Kyo said, multipitch vb's are nutworthy and very helpful in tuning. And to make a bank sound smoother you can mess with Velocity and also preutterance.

As for like actual pitch bends, stay away from a la carte unless you absolutely know what you're doing because it makes the utau sound drunk and it gets rid of any nice pitchbends you have done beforehand. So either use it before tuning or not at all. But setting your own pitchbends, it really helps to listen to human versions of vocaloid covers because then you can make pitchbends that echo the utaite cover or sound similar.

You don't need any extravagant bullshit or anything, seriously, as long as it sounds natural or nice, then??? who cares how your tuning looks.

It's important to mess with pitchbends and hear what it spunds like if you pull one way down or way up because those can impact your cover greatly.

Here's a lil thing i tuned:

and this one
https://clyp.it/hrt31jvj

both using that same technique.

And if there are little covers using actual Utaites (like Rera....) then you have to be really familiar with the song and experiment to see what sounds goon.

Add more control points to your pitchbends, take some away, put some up higher, put some down lower, stretch them out, make them narrower.

Modulation is also a goodie. People are always like "Omg don't use it unless your kyaami" but tbh it's really not hard to use. I like to use 30-35 modulation and then lower it on the notes that sound gross and off tune after i'm done tuning!!

Lol that's true about A La Carte. Whenever I use it I generally don't keep it, for now I just test the waters to see what it can do. And this tuning sounds excellent!!

>=< Hate when people act as if there's only one Goddess of tuning,,, And, yeah, I was considering trying to manipulate pitch bends on my own and use modulation, saw some tutorials regarding that that can make an Utau pop, a lot of times I'm too dissatisfied with how a UST makes my Utau sound to plug, play, and bolt ;^^
 
Similar threads
Thread starter Title Forum Replies Date
thehatman Can't hear tuning/has no effect UtaHelp 2
thehatman About tuning plugins on utau UtaHelp 2
blackrockshootersbff Help on tuning ??!! UtaHelp 0
Emilia Tuning advice UtaHelp 2

Similar threads