CeVIO Tohoku Kiritan CeVIO AI Announced
- Thread starter Rylitah
- Start date
I do not see a good reason why they would put Itako on for the same reasons Kiritan being on CeVIO confuses me. She is already on a AI synth(technically a hybrid synth since it has an editor), so unless CeVIO AI is massively different from Neutrino, I feel like she will only sound slightly better.
If I had to choose a Tohoku sister, I would have to go with Zunko. She may have a V4, but she does not have a AI-synth bank, so there would be a good reason to make a AI-bank for her.
If I had to choose a Tohoku sister, I would have to go with Zunko. She may have a V4, but she does not have a AI-synth bank, so there would be a good reason to make a AI-bank for her.
I can see it happening, only cuz Cevio AI Kiritan happened and ( it appears that ) it's easier to re-purpose Neutrino banks into Cevio AI.
But personally I'm with Stardust, I'd consider it pretty useless the same way I consider Neutrino Itako useless, she's pretty much identical to Kiritan ( I once sent my gf sample where I switched between Kiritan and Itako and she couldn't at all tell that I did that ). Unless Cevio AI fixes that massive issue I see no purpose in her being on there either.
But personally I'm with Stardust, I'd consider it pretty useless the same way I consider Neutrino Itako useless, she's pretty much identical to Kiritan ( I once sent my gf sample where I switched between Kiritan and Itako and she couldn't at all tell that I did that ). Unless Cevio AI fixes that massive issue I see no purpose in her being on there either.
I'm hoping they'll consider Zunko for CeVIO AI. Vocaloid is notoriously hit or miss with breathy vocals, so the change might actually benefit her.
I largely agree with most points Uncreepy made regarding this. I also think it's largely down to monetization, and the fact that CeVIO AI will probably be much more user friendly than Neutrino (I know an editor exists, but that's beside the point) even if it delivers high-quality results. People have posted sound clips of when Neutrino messes up and I imagine CeVIO won't have any of that. With the acapella clips of Kaf we heard, we know she'll be of higher quality too.
anyways, consider me perched.
I largely agree with most points Uncreepy made regarding this. I also think it's largely down to monetization, and the fact that CeVIO AI will probably be much more user friendly than Neutrino (I know an editor exists, but that's beside the point) even if it delivers high-quality results. People have posted sound clips of when Neutrino messes up and I imagine CeVIO won't have any of that. With the acapella clips of Kaf we heard, we know she'll be of higher quality too.
anyways, consider me perched.
Yeah, even Japanese users said they couldn't tell the difference between Itako and Kiritan's AI voices just by hearing.
I think if I remember right, Itako's AI voice data was crowdfunded when they did that lip sync medical technology. Zunko must already have AI data, because they made her have the Voidol voice which required some.
This is kinda tinfoil hatty, but doesn't it seem like AHS and CeVIO are having a hand in rounding up popular characters? It kind of seems like they are plotting something.
I guess one of my questions is about Kiritan and Itako's UTAU banks. Will they take them down if they end up being commerical at all? Does that mean that the UTAU vbs will be private after that?
Regarding the press release statement, I'd just like to bring this part to light: "CeVIO AI" not only can realistically reproduce human singing and speaking voices, but also allows you to freely edit pitch patterns, timings, etc. with an easy-to-use GUI, and there is a possibility of new voice creation that is different from the past."
Hopefully, this can satiate some peoples' curiosity since there was clearly some worry about whether or not you'd have the ability to tune at all.
Hopefully, this can satiate some peoples' curiosity since there was clearly some worry about whether or not you'd have the ability to tune at all.
With the stuff we've heard so far, I really believe CeVIO now has the opportunity to become one of the best in the English synth game, at least in terms of creating an actually natural-sounding one. AI is clearly capable of capturing nuances in pronunciation and whatnot, and I think that method could be perfect for synthesizing English since singing is so much different from speaking (which is what most Engloids sound like they're doing when compared to IA and Eleanor Forte). I'm not saying Engloids sound bad, per se, but they feel like they're missing something that the two previous captured. Maybe this is me just being desperate for an IA Eng AI bank, but I truly do believe CeVIO can do something special regarding the synthesis of English.
I doubt it. The UTAU banks were also crowdfunded, i.e people paid money to make them happen. I imagine said people wouldn't be very happy if they were to be taken down, plus SSS (the Tohokus' company) seems to be very proud of the fact that they've expanded to so many synths.
NEUTRINO is still free, and Kiritan and Itako are both on it, and nothing has happened to their UTAU banks.
The only way they'd be taken down is if the CeVIO contract demands them to be (why Makune Hachi, for example, was taken down -- she couldn't exist at the same time as Ruby due to her contract, though apparently the exact contract differed for every English Vocaloid) - but if this were the case, I'm sure there would have been an announcement already giving everyone a heads up.
So I don't know if this helps but we already have a synth using the same kiritan samples NNSVS
( Listen to 2020/05/24 Haru Ga Kita (4) by r9y9 on #SoundCloud
2020/05/24 Haru Ga Kita (4) )
and she sounds different from neutrino. You have to remember that the engine colors the voice just look at miku nt she uses the same source and sounds extremely different than vocaloid miku. I think it will also fix the itako and kiritan sounding similar.
( Listen to 2020/05/24 Haru Ga Kita (4) by r9y9 on #SoundCloud
2020/05/24 Haru Ga Kita (4) )
and she sounds different from neutrino. You have to remember that the engine colors the voice just look at miku nt she uses the same source and sounds extremely different than vocaloid miku. I think it will also fix the itako and kiritan sounding similar.
Release date Feburary 12.
Standalone Physical: 10800 yen
Starter Pack Physical: 19000 yen
Standalone Download: 9800 yen
Starter Pack Download: 18000 yen
Demo songs are by OSTER Project, maya, and Masakiritan-P (PchanZ).
Oh, and AHS is selling Genbu today now too (download only).
Standalone Physical: 10800 yen
Starter Pack Physical: 19000 yen
Standalone Download: 9800 yen
Starter Pack Download: 18000 yen
Demo songs are by OSTER Project, maya, and Masakiritan-P (PchanZ).
Oh, and AHS is selling Genbu today now too (download only).
Last edited:
I adore the way she sounds... but honestly? She doesn't strike me as sounding particularly different than her Neutrino bank (especially that last demo). And considering her Neutrino is free, it makes me wonder if that was really a good decision. I apologize if I sound too pessimistic, I just... are people gonna dish out a minimum of $100 to get results they can get for free with just a bit of extra elbow grease?
Hmm... whatever the case, I wish her the best and hope she does well!
Hmm... whatever the case, I wish her the best and hope she does well!
Last edited:
I still feel weird about her free singing dataset being used....for commercial product, on the more expensive end at that, doesn't sit right with me but ah well, best of luck to her.
Not the biggest fan of her Idol Neutrino voice (she does sound really good, just not my taste), but I will say she sounds slightly more stable here in CevioAI than she did in Neutrino, and being able to edit the voice in an actual dedicated editor is a massive game changer by itself. What confuses and bothers me here is; what is the end game for SSS / SHACHI in regards to future Neutrino voicebanks for the rest of the Tohoku sisters, and also Neutrino development in general? Unless Neutrino is just a testing ground for future sound source development, which would make sense since it is free for anyone to download and use. Its just weird to me that Zunko is being funded for a Neutrino release right now and not a Cevio one..
Side tangent; my vocal synth pet peeve is when developers put out wildly inconsistent sounding products for their characters across different synthesizers, while claiming that they are the exact same unchanged character. Yukari Rei and Xingchen Minus are at least conceptually different from Yukari Jun and Xingchen V4. I will miss Kiritan's aloof UTAU voice. Oh well, this still isn't nearly as bad as 1st place giving IA 2 Japanese V3 Voicebanks, 1 English Cevio on a program with barely any English support, 1 Japanese Cevio sister that doesn't mesh well with IA V3, and 1 Cevio talk voice that isn't even voiced by Lia herself and sounds absolutely nothing like IA..
Side tangent; my vocal synth pet peeve is when developers put out wildly inconsistent sounding products for their characters across different synthesizers, while claiming that they are the exact same unchanged character. Yukari Rei and Xingchen Minus are at least conceptually different from Yukari Jun and Xingchen V4. I will miss Kiritan's aloof UTAU voice. Oh well, this still isn't nearly as bad as 1st place giving IA 2 Japanese V3 Voicebanks, 1 English Cevio on a program with barely any English support, 1 Japanese Cevio sister that doesn't mesh well with IA V3, and 1 Cevio talk voice that isn't even voiced by Lia herself and sounds absolutely nothing like IA..
Last edited:
She have better singing control, breathing and change of pitchs between words sounds more believable, she sounds more loli now. But then Kiritan have a loli look despite being the older sister of Itako.
The quality can still vary depending on the producer cause the last demo sounds..raw? If you want it raw and noisy then better to just stick to Neutrino, which btw makes Kiritan sound like Sasara on some occasions.
It's not a huge leap. more like... 15%-20%? It's still an improvement and I'll take it as a positive thing.
The quality can still vary depending on the producer cause the last demo sounds..raw? If you want it raw and noisy then better to just stick to Neutrino, which btw makes Kiritan sound like Sasara on some occasions.
It's not a huge leap. more like... 15%-20%? It's still an improvement and I'll take it as a positive thing.
She sounds a little nicer in my opinion it doesn't have that neutrino sound but that slight cevio buzz. I wonder if we'll get video with her neutrino and her cevio ai to truly hear a difference. Do we know when cevio ai is going to get released.?
Kiritan's official age is 11. Itako is 19.
It releases with Yukari later this month, I don't know/haven't heard of the editor being sold on its own though.
My mistake, She's the YOUNGER sister of Itako. Thanks for that.
There is this Japanese article by DTM where they review Kiritan's Cevio AI bank. I can't read it though but the author tried a raw rendering on both engines to compare the 2.
AI歌声合成最前線。NEUTRINOとはどう違うの？2月12日発売が決まったCeVIO AI対応の東北きりたんを一足早く試してみた｜藤本健の “DTMステーション”
Neutrino!Kiritan
CevioAI!Kiritan
WHY DOES KIRITAN STILL SOUND LIKE SASARA??!! Is it cause the samples used is the same as the one in the Neutrino engine? If Himika Akaneya re-recorded the samples, Would the final product sound different or the same?
He also explained the matter of the producer's preference since he can change the voice quality parameters to deeper or squeekier with the option of more noise. This is redundant as it was already explained before.
Last edited:
I imagine that, because their schtick is that the girls are available on many engines, they will use the data from the Zunko Neutrino crowdfunding to eventually put Zunko on CeVIO AI. To me, it's basically the AI equivalent to them being available on both VOCALOID and UTAU. I know because we are western fans we don't pay as much attention to the crowdfunding, but if they do stuff like surpass the goal, they get other perks through stretch goals. I think the more money the Touhoku family gets, the more likely it is for western fans to get voice banks in English down the line and maybe they'll eventually do stuff with the other characters in Zunko's world if they run out of companies to collab with, haha.
I think that with a lot of classic vocal synth girls (Yukari, flower, IA, etc) appearing on CeVIO AI, it seems convenient to have them all in one editor, too. (Sorry this post is written poorly, very tired.)
Edit: I think it's possible, too, that the Touhokus might get CeVIO AI Talk voices, don't you think? And then the song and talk voices can be in the same editor.