Before I start recoding my banks..

Linkmaster647

Momo's Minion
I've informed myself of the basics and all, I found a reclist, I have Oremo downloaded and all, have a good idea of how to record a bank (going for a VCV) but still have some questions...
1: How bad is it if I don't have a power/soft versions of the bank? (Like would the intensity of a song lost because of that, for instance, how bad would be to use a normal bank for covering a song like Lost one's weeping?)
2: Same question but for monopitch, I know multipitch makes things smoother but HOW MUCH does it help?
3: as for power/soft bank versions... is there a way to put some sort of filter on the track and still make it sound good, otherwise, tips for recoding thoses bank types?
4: (and lastly) what pitch to record on? I know if I end up doing multipitch I'll go with 3 (one high, one middle and one low) but at least for the middle one... what is the most "middle note" for a pitch?
 

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
If you're starting to make voicebanks for the very first time, don't worry too much about having big grand plans or making voicebanks as complicated as long-experienced users. In fact, you may want to start with a basic, single pitch CV Japanese voicebank, so that you can make sure that all the technical setup is correct and UTAU works properly without errors.

1: How bad is it if I don't have a power/soft versions of the bank? (Like would the intensity of a song lost because of that, for instance, how bad would be to use a normal bank for covering a song like Lost one's weeping?)

Voicebanks only need to have as many expressions as you want to make. Having extra ones isn't a requirement. In fact, if you intend the voicebank to suit one specific music genre, you'd only have to make that one tone, and it's not "missing" anything if you don't have extras.

2: Same question but for monopitch, I know multipitch makes things smoother but HOW MUCH does it help?

Multipitch doesn't make the vocals sound smoother or more connected, the reclist formatting and otoing do. What multipitch achieves is changing tone throughout the vocal range, from low to high notes. A single pitch voicebank will already have a decent usable range, and it'll generally sound better on notes higher than the original pitch than lower. That means if you're only recording one pitch, pick one at the middle-low part of your range (not the lowest part). If you already like the results, great! You can add additional pitches later if you want.

3: as for power/soft bank versions... is there a way to put some sort of filter on the track and still make it sound good, otherwise, tips for recoding thoses bank types?

You can use the Mt (tension) flag with moresampler and the default resampler in OpenUtau. By default the value is 0, and it ranges from Mt-100 to Mt+100. Lower values make vocals sound softer and higher values make it sound stronger.

4: (and lastly) what pitch to record on? I know if I end up doing multipitch I'll go with 3 (one high, one middle and one low) but at least for the middle one... what is the most "middle note" for a pitch?

I touched on this topic in the previous question, where you'd want to pick pitches at the middle-low range of where they're used. As for exact specific pitches, that will depend on your own personal singing range, and the tone of voice you have at different notes. For example, if you have very consistent tone, then you don't have to record many pitches, and they can be spaced apart evenly. But if you were to have a very dynamic changing tone (like from chest voice to head voice to falsetto) then you might want to have more pitches and space them around where your tone changes are.

Of course, that all assumes you want to record a voicebank that perfectly captures your natural voice. Quite a lot of voicebanks are for characters and have voice acting! So these sorts of tonal changes also depend on the sort of character you want to depict and the sort of voice you think they should have.
 

Linkmaster647

Momo's Minion
Thread starter
If you're starting to make voicebanks for the very first time, don't worry too much about having big grand plans or making voicebanks as complicated as long-experienced users. In fact, you may want to start with a basic, single pitch CV Japanese voicebank, so that you can make sure that all the technical setup is correct and UTAU works properly without errors.

Well I know how VCV works it doesn't seem hard as for example on the reclist I found
ke_ke_ku_ko_ke_ko_ko
I know that VCV would just makes
eke, eku, uko, oke, eko, oko
So on that end, I'm pretty confident as to what I'm getting about

Multipitch doesn't make the vocals sound smoother or more connected, the reclist formatting and otoing do. What multipitch achieves is changing tone throughout the vocal range, from low to high notes. A single pitch voicebank will already have a decent usable range, and it'll generally sound better on notes higher than the original pitch than lower. That means if you're only recording one pitch, pick one at the middle-low part of your range (not the lowest part). If you already like the results, great! You can add additional pitches later if you want.

On that end note, I was just worried that the voicebank wouldn't sound as good when getting at high notes or low notes, I did fiddle around Rouon Aro voicebank looking at ust I found and I believe on the song KING, Rouon voice sounded well higher pitched that I was used to so it looked wierd to me

You can use the Mt (tension) flag with moresampler and the default resampler in OpenUtau. By default the value is 0, and it ranges from Mt-100 to Mt+100. Lower values make vocals sound softer and higher values make it sound stronger.

So that can just be an alternate way to make power and soft? Will keep that in note, if I'm getting correctly the resampler is one of the 2 tools you can setup when you bootup an ust right?

Of course, that all assumes you want to record a voicebank that perfectly captures your natural voice. Quite a lot of voicebanks are for characters and have voice acting! So these sorts of tonal changes also depend on the sort of character you want to depict and the sort of voice you think they should have.

Lastly, actually fun fact for you (and me getting easy on me for 1 of my 2 voicebanks, because yes, I plan to have 2 characters) but the first character I'll a voice bank of is actually the one you see as my profile picture here, and this character is basically supposed to be me IRL (but as an animal) anyways, that just means that for that first voicebank, I won't be voice acting at all, it'll be my own natural voice since the character is supposed to be me!


Also already just this helped me quite a bit, at first I was scared of the amount of work I'd have to do, thinking that I'll need to make like 9 voice banks just to have 3 different banks (for power, normal and soft) all at tritonic...
At least knowing this just means... I could make something good sounding with just... a single voicebank right?
 
  • Like
Reactions: Kiyoteru

Linkmaster647

Momo's Minion
Thread starter
Right I forgot one thing! I mentionned a second character and um... for a voicebank would it still sound good if I put a sort of "robotic" filter on it? (The character is a protogen, if you don't know what that is, basically a half robotic, half animal sort of character, hence why the robotic filter I might put on his bank)
 

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
Robot characters are a common trope in UTAU and they're still given natural voicebanks, so I don't think such editing is strictly necessary in order for people to understand that it's meant to be a robot character.

However, if you do still want to edit samples because you like the sound, keep in mind that UTAU itself is going to add another layer of processing afterwards, and you may not get the results you expected. Generally I would advise recording samples as cleanly as possible to avoid the need for editing (eg. record in a quiet room so you don't have to use noise removal). You can add effects on a song-by-song basis while you're mixing the final vocal tracks. However, mixing effects will be up to the end user of the voicebank, and if you distribute it to the public, that means other people will be responsible for applying those mixing effects. If you edit the raw samples, this will force all users to start with the same effects already applied.
 
  • Like
Reactions: Linkmaster647

Linkmaster647

Momo's Minion
Thread starter
Robot characters are a common trope in UTAU and they're still given natural voicebanks, so I don't think such editing is strictly necessary in order for people to understand that it's meant to be a robot character.

However, if you do still want to edit samples because you like the sound, keep in mind that UTAU itself is going to add another layer of processing afterwards, and you may not get the results you expected. Generally I would advise recording samples as cleanly as possible to avoid the need for editing (eg. record in a quiet room so you don't have to use noise removal). You can add effects on a song-by-song basis while you're mixing the final vocal tracks. However, mixing effects will be up to the end user of the voicebank, and if you distribute it to the public, that means other people will be responsible for applying those mixing effects. If you edit the raw samples, this will force all users to start with the same effects already applied.

I see alright, as for sound quality, I made sure to order some top notch mic (I bought the Yeti brand) so I think for audio quality I'll be fine!
 

Linkmaster647

Momo's Minion
Thread starter
I'm trying the moresampler to check how the mt works but Utau says it fails to generate a wav file...
 

Similar threads