How do you mix your vocals before importing them into the UTAU program?

[TakeruInudo]

I'm just curious as to how everyone mixes theirs? I know if I want a nice clear bank, what I do is after I'm done recording and fix all my words up I'll noise remove all the sound (to the point where you can't hear anything, not just up to barely hearing like most voicebanks Ive heard/seen) then I EQ and compress them. And if I want a more softer sound then I add some bass/treble to it.

so what about you guys? how do you all mix your vocals and for what effect before you put them into the utau program??

 

[Meta]

I don't do anything to them I just use the raw samples. I've got a nice enough set up where I don't really need to doctor them for them to sound decent.

 

[Karma Drak-pa]

I only began to do it when creating some prototypes some months ago, but i simply remove the background noise (I did it besore with Edison and now with RX4) and I remove the mouth noises with a declicker. Also, if I think an unvoiced consonant is a bit too long, I shorten it (but I have to be really careful).
I actually do that for all my other recordings too (except the consonant shortening), such as my (rare now) tutorials’ audio track.

 

[Kiyoteru]

My recording setup is pretty decent, so I've never really felt a need to edit my samples. For certain tones I know they could sound better with some EQ, but I usually decide that it'll come in mixing the final vocals rather than before it goes through UTAU.

 

[Lorreine Geralde]

All I remembered was...I remove noise background and "click and pop" on Wavepad and use some mastering settings on Adobe Audition (trial version) for the whole recording. These are methods I used when creating my UTAU's latest voicebanks (Kaiserine Sympherianne's "REVIVAL" and "REVIVAL SE" V.

 

[MillyAqualine]

I rarely edit my VBs before putting them inside of UTAU

The only edits I've done so far were : a pitchchange (Kei Den and another old bank to make them sound actually male and more mature), noise removal a few times (mostly on Kei due to his whispery/breathy voice but he wasn't all alone) and kerovee (for a voicebank which is an experiment of "How UTAU does react to this effect..." )

I've also used EQ+noise removal combo' on a few phonems from another whispery voicebank, and another time I wanted to test a very slight echo (and I emphasize on the "very slight" ) to see how it'd react

But yeah, generally I use the raw samples directly

 

[Mako]

I've only done some minor pitch changes in the past and some EQ'ing, noise removing and compressing/amplifying for Tori's ACT2 bank. The pitch changing sounded alright at the time, so did EQ'ing and noise removing give more clarity and quality to his ACT2 bank, but I wouldn't recommend amplifying the vocals. Or not amplify them like I did, a lot. It tends to sound too loud and/or gets grainy or something i dont really know how to explain it sorry owo

Maybe you should stick with the raw samples, like said earlier. Mix the samples only if necessary. owo

 

[TakeruInudo]

I've only done some minor pitch changes in the past and some EQ'ing, noise removing and compressing/amplifying for Tori's ACT2 bank. The pitch changing sounded alright at the time, so did EQ'ing and noise removing give more clarity and quality to his ACT2 bank, but I wouldn't recommend amplifying the vocals. Or not amplify them like I did, a lot. It tends to sound too loud and/or gets grainy or something i dont really know how to explain it sorry owo

Maybe you should stick with the raw samples, like said earlier. Mix the samples only if necessary. owo

I usualy try to keep the vocals between a -5Db and a -10Db because of the grainy-ness and other things like you said, though my house echos a lot so leaving them unedited makes them sound so muffled >.<

 

[Karma Drak-pa]

I usualy try to keep the vocals between a -5Db and a -10Db because of the grainy-ness and other things like you said, though my house echos a lot so leaving them unedited makes them sound so muffled >.<

Personally, I don’t recommend to have louder samples than -6dB, because if you double the volume of a -6dB sample, you get a 0dB sample; that’s what you get when you turn the volume of a note from 100 to 200. If you have louder samples, you can get saturation in your vocals, and no one wants that, except for some type of songs (but I think it’s always better to have vocals without saturation and add it in your mix later, so you can control it)

 

[TakeruInudo]

Personally, I don’t recommend to have louder samples than -6dB, because if you double the volume of a -6dB sample, you get a 0dB sample; that’s what you get when you turn the volume of a note from 100 to 200. If you have louder samples, you can get saturation in your vocals, and no one wants that, except for some type of songs (but I think it’s always better to have vocals without saturation and add it in your mix later, so you can control it)

I'll remember to keep that in mind the next time I make another VB I'm finishing up my latest one now (I like to experiment with my VB and the affects they will have when you mix them a certain way when you go to put them into the utau program), but that's on a small hiatus at the moment since I got flooded with other work.

 

[AmeKinoko]

I just record with oremo and don't do anything fancy, really. I have a pretty nice microphone, so things usually sound fine.

 

[Chuo]

I sometimes use eq when recording the original samples but otherwise I don't touch them
One time I put a small amount reverb on my low pitch cv bank and it made it sound very cool and robotic, so it might be worth trying some of the effects

 

[Hentai]

So for Otakune Weeaboo we experimented with something that we would later call the "Modified Weeaboo Method".

When Otakune was recorded his raw samples were heavily edited in Melodyne. Using Melodyne all his samples were made pitch perfect, all vocal drift was removed, and all voice modulation was flattened. The above was executed in an extreme method with the goal of making Otakune sound as robotic as possible. In this matter we were highly successful, Otakune sounded extremely robotic but it had the unintended side effect of making Otakune exceptionally clear and with less variation in the voice, combined with clean samples, the resamplers love him.

That being said typically your goal is to not sound completely robotic, thus later experiments in the "Modified Weeaboo Method" would focus on pitch perfect, less drift samples, without editing the modulation too much as to preserve as much of the "human" qualities of a voicebank while still improving clarity. Its a bit of a slider, takes a bit of experimentation, and doesn't like OREIMO but was a pretty fun experiment while it lasted.

 

[TakeruInudo]

So for Otakune Weeaboo we experimented with something that we would later call the "Modified Weeaboo Method".

When Otakune was recorded his raw samples were heavily edited in Melodyne. Using Melodyne all his samples were made pitch perfect, all vocal drift was removed, and all voice modulation was flattened. The above was executed in an extreme method with the goal of making Otakune sound as robotic as possible. In this matter we were highly successful, Otakune sounded extremely robotic but it had the unintended side effect of making Otakune exceptionally clear and with less variation in the voice, combined with clean samples, the resamplers love him.

That being said typically your goal is to not sound completely robotic, thus later experiments in the "Modified Weeaboo Method" would focus on pitch perfect, less drift samples, without editing the modulation too much as to preserve as much of the "human" qualities of a voicebank while still improving clarity. Its a bit of a slider, takes a bit of experimentation, and doesn't like OREIMO but was a pretty fun experiment while it lasted.

That's neat.