CeVIO Kafu announced for CeVIO AI (Early 2021)

[morrysillusion]

as noted by people pitching the samples up and down, im wonder how much the Type B voice would even avoid people making her sound like Kaf? if all you have to do is just...pitch her down...? even without a GEN parameter it takes no work at all to pitch it down, as shown previously. it would make sense if they encouraged Kaf to have voiced the Cevio AI voice in a way like a character voice actor would, with a different accent/tone/annunciation but it seems like its just her. i can get the sentiment of not wanting her to be identical, but i think they shouldve approached the voicing in a different way if the wanted to do that, as i wouldve preferred to not see the options as "sounds like Kaf" and "is moe"... but i guess im just personally tired of moe voices.

 

[___]

I'm not against another cutesy voice as much as I am displeased with...well...the lack of direction? To me it just seems as if they weren't even expecting Kafu to come out this realistic and then went "...Oh shit" and pitched her up. I guess what I don't like is that type B comes across as afterthought more than anything. I wouldn't be as/at all displeased if they actually asked Kaf herself to put on a cuter tone for the vb.

 

[morrysillusion]

To me it just seems as if they weren't even expecting Kafu to come out this realistic and then went "...Oh shit" and pitched her up [...] I wouldn't be as/at all displeased if they actually asked Kaf herself to put on a cuter tone for the vb.

this is exactly what i was starting to feel-- the first sample they gave just did not sound like the synth's natural range/pitch and i wonder if comments from fans about making Kaf into a synthesized voice made them show it off that way.. it doesnt sound like the bank was recorded at that pitch at all and im quite sure he range would be super limited if that was in fact her default area as she already sounds near strained. at this point im surprised they asking what people want to hear bc like, i assume the bank is done being recorded and with its release date coming up, what would they change? are they just going to say her range is around that high sample and expect people to use her that way? if they had just recorded her with the moe tone in the first place, it would probably sound better at least. its obvious something is up, when people can pitch down those samples and they sound just like Kaf's normal range.

 

[xuu]

I better see every single person voting for Type A. B just being a raised formant version of A means it does effectively nothing but lower quality and C is well... a waste of money on an AI synth. If they want a B or C type voice then she should have been developed for CeVIO Pro instead. A is definitely the original vocal and they almost certainly created that then had to post the demo using type B to avoid backlash from fans of Kaf, but then it created even more backlash because it sounded well... like that. Glad to know that the sample was edited though.

Alternatively, if they can do it this easily... just release Type A and Type B.

 

[uncreepy]

this is exactly what i was starting to feel-- the first sample they gave just did not sound like the synth's natural range/pitch and i wonder if comments from fans about making Kaf into a synthesized voice made them show it off that way.. it doesnt sound like the bank was recorded at that pitch at all and im quite sure he range would be super limited if that was in fact her default area as she already sounds near strained. at this point im surprised they asking what people want to hear bc like, i assume the bank is done being recorded and with its release date coming up, what would they change? are they just going to say her range is around that high sample and expect people to use her that way? if they had just recorded her with the moe tone in the first place, it would probably sound better at least. its obvious something is up, when people can pitch down those samples and they sound just like Kaf's normal range.

The thing is that the original announcement/demo song was a complete surprise. So fans couldn't have given feedback on it which made them decide to pitch it higher. The team must've realized it at the last minute like you said, by on their own. It seems weird that they're like "Kaf herself will read your feedback!" Like... shouldn't she of all people know if she wants people using her exact voice or if she would be more comfortable with a pitched up version?

Also, the release date is this year, so it seems really weird to change it at the last second. Are there even going to be boxes for people to buy or will this be digital voice bank downloads only?

The more we think about this/discuss it, the weirder it gets.

 

[lIlI]

Both a and b seems popular. I honestly would of wanted a D the had her forment lower. I hope A wins so we can have the most natural bank. I could see all of them being appends tbh. I feel like c kinda ruins the purpose of ai.

Ask and ye shall receive, Type A pitched down 2 semitones just for fun:

https://clyp.it/sxkvawvc

 

[Prism]

That's way more pleasant to my ears.

 

[YOYo_MAMA]

If they wanted to avoid the backlash from KAF fans, they probably should have designed KAFU to be a complete separate character from KAF, or change the name so it won't be identified with KAF. KAFU looks so much like KAF's avatar design that it's impossible to not identify the OG from it, and that is their intention.

This backlash from fans remind me of the reaction of people when it was announced there are more than one Kizuna AI. Fan's will always say that there's only one Kizuna AI and not those "copycats"

Since Kaf already have an identified image in Japan that fan can associate it, making a VB from someone who is already famous and known will have a loyal "purist" fanbase. They will be against it to protect their idol's vocal virtue. Oversees fan's don't have much to complain since they never heard of KAF so they can't make the connection that easily. I've never heard of KAF until Cevio announced this collaboration.

 

[Blue Of Mind]

I don't understand now why they didn't make Type A and B as append banks for Kafu... As in, Type A would be Kafu Natural and Type B would be Kafu Sweet, or something like that. I deliberately ignored Type C because it literally just sounds like Type B on AutoTune.

Honestly, this announcement is turning into a mess. Not a good look for CeVIO AI.

 

[uncreepy]

Honestly, this announcement is turning into a mess. Not a good look for CeVIO AI.

To be fair, Techno-Speech doesn't seem to have anything to do with this decision from Kaf's team.
But I have the feeling people are going to be a bit emotional about the other characters when we get to hear them.

 

[uncreepy]

I finally got around to filling out my answer to the survey. I basically just summarized the general opinion of the western fandom/why we think that way. As in most didn't know who Kaf is, but they liked her design/voice when they heard it and like TYPE A because it sounds like her and makes sense if it sounds like her because it's AI. Versus the Japanese users who are probably overwhelmingly just Kaf fans and not producers who want to protect her voice.

This is random/dumb, but even though I know who Kaf (花譜) is, the voice in my head keeps reading 可不(KAFU) as "FUKA" because "不可能 = fukanou = impossible". So if I ever accidentally write Fuka, you know why. (I've caught it when I do it, but one of these days, I might not notice/post it.)

 

[Prism]

I do find it interesting that this is the first singing synth that has this problem. We have celebrity voice banks in the past with bigger names I wonder why this one is having this dilemma

 

[uncreepy]

I think because Kaf's team decided to make it a problem, haha. It could've just been their problem, but they decided to make it *checks their Twitter follower amount* 6,000 people's problem.

 

[Prism]

Now thinking about it I wonder if that's why chika doesn't sound her voice provider

 

[Blue Of Mind]

I do find it interesting that this is the first singing synth that has this problem. We have celebrity voice banks in the past with bigger names I wonder why this one is having this dilemma

I think because Kaf's team decided to make it a problem, haha. It could've just been their problem, but they decided to make it *checks their Twitter follower amount* 6,000 people's problem.

I was gonna say if anyone complained about Gakupo sounding too close to GACKT, or IA sounding too close to Lia* or something along those lines.

*Personally, I don't think IA sounds anything like Lia at all, which annoyed the hell outta me as a Lia fan at first, but now I like it. I just consider the Vocaloid character and VB as really separate from the VP. I should probably take this to the Unpopular Opinions thread. :P

 

[YOYo_MAMA]

I do find it interesting that this is the first singing synth that has this problem. We have celebrity voice banks in the past with bigger names I wonder why this one is having this dilemma

I have a hunch it because of the Vtuber fanbase and how much publicity KAF's company of promoting this right now. This problem is mostly just in Japan where the KAFU fanbase is.

KAF is the 1st Vtuber that I've heard of who will have a Synch based of her, and is still active right now.
She's not going in hiatus like Lia (who was giving birth at the time) when IA was first released, even then the publicity of IA is not as immediate as KAFU. For Gakupo, the Vocaloid fanbase so sometimes turn him to a meme for the laughs. Even back then the Vocaloid fanbase was not as huge as it now.

Vtuber's are rising very quickly, more than the combined years it took for Vocaloid to become the subculture to it is now. First was Kizuna AI, then many more start popping up until Hololive became a thing.

Vtuber and Vocaloid fans both idolize an "anime" Waifu/Husbando with a voice and songs they can "interact" with, escape to, and they are fierce in defending it.

 

[Prism]

I think this issue best mirrors Piko he was a utaite they both have large internet fandoms and are semi professional. I do wonder if it's because the ai was so good. The past vocaloids have sounded a little bit different from their voice providers but kafu sounds real close to kaf in how she sings it's not flat pitch like vocaloid

 

[YOYo_MAMA]

I think this issue best mirrors Piko he was a utaite they both have large internet fandoms and are semi professional. I do wonder if it's because the ai was so good. The past vocaloids have sounded a little bit different from their voice providers but kafu sounds real close to kaf in how she sings it's not flat pitch like vocaloid

I forgot Piko's Voice provider was also in the Vocaloid scene as a utaite, maybe is cause Piko is not as mainstream as the Big "celebrity" now

Yoholoid also have a similar background but he was not as blown up as KAF's, and he's Voice Provider is Swedish.

(Flashbacks to when I thought Carameldamsen was Japanese, not Swedish)

 

[lIlI]

I do find it interesting that this is the first singing synth that has this problem. We have celebrity voice banks in the past with bigger names I wonder why this one is having this dilemma

This has come up before! It has a whole section on the Vocaloid wiki, and has been a concern since Miku. It's part of the reason Crypton switched direction from realism to voice acted vocals, and why Zero-G rarely released the names of their voice providers.

 

[uncreepy]

I think because Kaf's team decided to make it a problem, haha. It could've just been their problem, but they decided to make it *checks their Twitter follower amount* 6,000 people's problem.

Wow, I was right. Almost 6,000 people DID vote and TYPE A won.


Kaf will take the opinions into account and be the one to make the final decision, which will be announced tomorrow.