Utaus with different laguages

[HulderBulder]

Hello
I want to discuss on UTAU voicebanks that are made in certain languages. As we all know UTAU has the capability to make voicebank sing in different languages if the bank has the required sounds. As well that cv vc recording is most prefered for all languages that have ending consonants.
My question is, what is your view on banks that support only one language? Do you wish they supported other languages as well? When do you think they cover enough languages?
Also, do you think certain languages are harder to use (apart from not understanding it)?

 

[수연 <Suyeon>]

RE:

Banks with only one language

It's really the choice of the creator and/or voicer. I don't have a problem with a bank that only supports one language. Even with banks that have multiple languages, I'm bound to use one that easily converts to different formats (Japanese/Korean - pro: share the same phonemes even if pronunciation is different; con/semi pro: SeeU's hangul doesn't translate to UTAU, but her Japanese is accented and can be used in romaji).

Do you wish they supported other languages as well? When do you think they cover enough languages?

Fluency in a second/third language is nice to have, but I don't dwell on it, esp. if the voicer isn't fluent in the languages they're doing/mainly works with one language anyway. I think the amount of languages an utau should support should be based on a) the amount of usage it will see from the owner themselves in addition to other users, and b) whether there's a high demand for it. I guess... I have the same view point for utaus that I do for vocaloids: they take time to make - and in utau's case, unless you're a part of the Macne series, you're not being compensated in any way for the end product. I think that the time spent making an utau bank should be used to perfect the voices/languages that have already been established rather than make voices in other languages to satisfy others and coming out with lackluster results (*ahem* Crypton Future Media *ahem*). Is that to say that multi-language banks shouldn't be attempted? No. Rather, it's to say that they should be done one at a time and only after you can't improve a bank anymore.

Do you think certain languages are harder to use (apart from not understanding it)?

I think even English speakers can agree that English banks would be difficult to use without the "dictionary" of sorts created by Cdra and Chezzie. Not that English Vocaloid makes much more sense, but Vocaloid does show you the phonemes in addition to the lyrics, where in utau, you would have to go through the oto and listen to each sound until you found the one you wanted. Second hardest in terms of difficulty for a non-native speaker would probably be Korean (recording it more so than using it) due to it's having 3 different kinds of consonants: regular, double/tense, and aspirated. To a non-native speaker, they can all sound the same.

 

[MillyAqualine]

RE:

Honestly, it's not a big deal.... Sure it makes edits harder, but at the same times, I'm having more fun because it' like a tiny challenge (like Teto, Axel or any single-languaged bank I own, when I try to make them sing in French, English Spanish despite not having the phonems or all of them)

Then I'm trying to understand the person who created the bank; they just wanted to make a bank for themselves and for the fun, or they don't know foreign languages very well, or they don't feel the need to record languages they know. Of course, it'd be fun and great to have other languages but as I said, it's up to the author. And honestly, I won't push them for this as I enjoy their voicebanks the way they are~ ^^

 

[Zarsla]

RE:

For me it doesn't really matter how many languages a voicebank can do; it's how well it can do all those languages. For example Japanese banks with added English appendages I don't really like because when they do sing in english it's either muffled or I have to do a lot of converting to get to sound good (eg clear and understandable).

Also many UTAU users make separate banks for languages (e.g. they have a Japanese bank and an English bank but they're two different voicebanks). Which, I think, is in part because of the different systems used for different languages in terms of coding. For example in Japanese Romaji a,e,i,o,u make the aw, eh, ee, oh and ooh sounds but in english it's a,e,E,O,o not to mention the different recording styles for Japanese it's CV,CVVC and VCV but for English it's just CVVC.

So... if you happened to make an English and Japanese bank or reclist, you'd have to figure which system to use to cover both sounds the languages use and how to code it (e.g. do you use CV or CVVC or VCV+CVVC or what? And then do you encode it using Japanese Romaji and then add the rest for English or do you code in just English?), and then how to deal with conversion from one to the other (like if I get a CVVC English ust, how do I use a Japanese Romaji encoded voicebank to make it sing or how do a VCV hirgana encoded ust sing using a English encoded bank). And then if you deal with distributing your UTAU voicebank out, you'd also have to know how you can explain it to others (Do a Soundcloud podcast? A Youtube tutorial? Put in the readme of your UTAU download?).

 

[Fuutari Makku]

Personally , I made my own reclist (I called it semi-universal and it's obviously CVVC). It can speak italian and japanese well and english (a bit , but since I'm italian , my english pronunciation isn't that good , but I added english r phonemes). I made the reclist on my own , but it isn't stringed (in fact the voicebank has got like 1600 samples) Since I haven't got any base oto , and otoing bothers me a lot. But basically I think I can handle some spanish and french , too , with my voicebank , since I added the spanish j but I made it similar to french's r (like in Maika for Vocaloid 3). It's a bit uncomfortable , since you have to remember phonemes name (but I can do that) and also pronunciation isn't good for every language. I always make Let It Go in 25 languages to demonstrate it (but I've only released a demo and never finished it since I lost the file , but I'm remaking it). I had many problems...but once it's finished (oto is still a wip) I think it gives you good results (if you can use it).
edit: (just made another demo https://soundcloud.com/boukaroidomeikar/another-let-it-go-demo-yay )

 

[IrisFlower]

1. Banks with only one language

It is all up to the creator imo. People shouldn't feel the need to record a second language if they don't want to or just can't for whatever reason. If your UTAU can only sing in Japanese fine. 9/10 times people use UTAU for Japanese anyway. If it can only sing in your native tongue? Also fine. If people like your UTAU enough they will find a way to use it trust me.

2. Do you wish they supported other languages? When is it enough?

Not gunna lie...if the song has a lot of English in it I do wish for English support but again...I can get around it. I feel like once you have three languages it's kind of enough per se but if you wanna record more be my guest. Just maybe like...don't package it all in one bank cause you could end up with a lot of phonemes lol and it'll be a lot to download. I'd do separate banks for separate languages because of overlapping phonemes(same alias, different sound).

3. Are certain languages harder to use?

Speaking from my own experience, since my Iris has an English CVVC bank, I can say that the only thing that makes it "harder" to use than her Japanese is having to think about how certain words run together and blend into each other. Like "fit in"; in UTAU I'd be more inclined to put the phonemes together as "fi tin" instead, ya know? But that's really it. I have no problem understanding the phonemes(I do make mistakes but not to the degree that I'd classify using the bank as "hard") and quite frankly prefer using English CVVC to Japanese CV or VCV. Would I prefer it be like it is with Vocaloid where you just type the lyric and the phonemes are put together for you(most of the time...)? Mmm...yeah.

BUT even Vocaloid English has the whole "how should I blend these sounds together" thing. At least I do it with Luka. Maybe not to the degree of UTAU but it's there. There are times where I take out ending consonants in words because the next word starts with that same consonant and she'll kind of "stutter" if I don't. And I'm pretty sure she is missing some transition sounds that I have to find a way to input manually...

Now, overall? I'd say that yes most other languages, especially if they're structured like English, are harder to use compared to Japanese. But I wouldn't really call it harder either...more like time consuming.

 

[Fuutari Makku]

You're right , I don't really find hard using English CVVC , but I can't use a CVVC (if it isn't mine) bank decently without editing oto.ini. It's kinda weird. The most evident phonemes problem I had with my bank is the r problem. There are 3 different r in my bank. So I called japanese/italian/spanish' s r simply r , the french one (or spanish j) x , and the english one 2 (since it's a bit similar to R shape to me). I think that one more problem is the pronunciation , as I said before...I don't really know if pronunciation sounds good for everly language in the demo (and I think it isn't much good for english) , I can only say for sure if pronunciation is good in italian (and maybe japanese?). But probabily there are even more r. The biggest problem I've got now is that , since my UTAU's act 1' s quality sucked , I deleted download link from the wikia and since I haven't finished oto (cause it's reaaaaaaaaaaaaally long) I can't release it. This may be really OFF TOPIC , but I should record another CV and I found a way to make a..."Fake VCV" (?) that consists in adding only six samples and vowel transitions to a normal CV bank.