Other Cryptonloid voicebank updates, collabs, & concert news (crypton_wat Twitter translations)

[uncreepy]

Because it's almost Meiko's 15th anniversary, we're in the middle of work, but we are going to cooperate and liven the mood*.

Even today we adjusted two plans that will be exposed by SEGA and MEIKO. Please wait a while longer for public details. Yoroshiku onegaishimasu.**

* I dunno how to translate this properly, but it's basically "cooperate and enliven", as in lift the mood/spirits/atmosphere.
** "I look forward to working together / please do me this favor" type meaning.

(Edit: I accidentally bumped my tinfoil hat theory about possible new designs being revealed at the bottom of p14, go back and check that out before reading this post!)

 

[peaches2217]

I ADORE these designs! I especially love Luka's redesign. I loved her flowy V4 skirt, but I always thought her headband/headphones looked a bit too bulky, and I adore her V2 top, so I'm very glad they brought it back.

I'm just... SO glad Meiko might get another update. She is my QUEEN and I NEED an update for her.

 

[uncreepy]

I'm just... SO glad Meiko might get another update. She is my QUEEN and I NEED an update for her.

It's not "might", she IS getting an update. : D All the Cryptonloids are. We won't know the release date til the end of the year, though.

 

[Blue Of Mind]

I'm glad to know that Meiko and Kaito are getting updated at least - they got skipped over in V4, and I think their English banks could be improved.

I can see that the (possible) redesigns are incorporating elements of all the character designs that have appeared so far, perhaps as a way to placate fans of the different designs? Or maybe it represents the fresh start away from the Vocaloid engine, so all the previous designs have been merged together for the start of the Piapro era? Or maybe I'm thinking too deeply about the second interpretation. I'm digging the Meiko design the most, because it's a happy medium between the V1 and V3 designs.

 

[peaches2217]

I'm digging the Meiko design the most, because it's a happy medium between the V1 and V3 designs.

The one and only thing I'm not a fan of is the black necklace/collar. It looks a little odd, but I feel like, if I see another picture of it, it'll grow on me. The rest of it I adore, and I can absolutely see this as intentional on Crypton's part - they took the most well-received parts of each version and smushed 'em together into some very slick-looking, cohesive designs. Whether or not that's the case, I think it's a very good move on their part.

 

[Prism]

Just had the idea could one use cherry pie on vocaloid or piapro to get English vocaloids closer to the Japanese version I'm thinking Len English

 

[uncreepy]

Just had the idea could one use cherry pie on vocaloid or piapro to get English vocaloids closer to the Japanese version I'm thinking Len English

We don't know if Cherry Pie will work with non-Cryptonloids. (Someone on Twitter said they asked Crypton staff during Magical Mirai and they said it was only Cryptonloids, but we can't be 100% sure if this is accurate hearsay.) But based on the demo, they used Miku to coat an English voice with her voice and it resulted in good-sounding English.

See 1:23 at the video (It's a female voice with the "Conversion to CV01" setting, aka Miku)

(Gosh, I am sick of having to watch this same old demo from March over and over.)

Unfortunately, we haven't gotten to hear any other English clips this entire 2-ish year span the thread's been following the Appends/next generation Piapro.

Your comment got me thinking, though. I remembered back when they had a presentation in June, the options for voice conversion were:


Starting under "Vocal Fry":
Male to CV01 (Miku)
Male to CV02 R (Rin)
Male to CV02 L (Len)

In the demo video I linked to, the setting was for CV01 (Miku). However, it doesn't appear there are different settings in order to access their English or Japanese. I think you just pick the character and can speak any language at Cherry Pie and it converts the voice (also going back to how Wat said their goal was "multi-lingualization"). (See my "What is a vocoder?" explanation: Cryptonloid voicebank updates, collabs, & concert news (crypton_wat Twitter translations) .) I think it would be like Voidol, where I was able to speak both English and Japanese and it had the voice converted to the character's (Here clips at this thread: Real-time voice conversion software aimed at VTubers called "Voidol" ).

But I don't really understand how using Cherry Pie (the vocoder) would be linked to the actual tuning part in the new Piapro, it's most likely going to be like an updated version of VocaListener where it detects the start/end of words and you can enter phonemes over the blank music notes, but I don't know. (Especially since AIST, who made VocaListener, is now working with Crypton.) (Maybe I should make a post about how VocaListener works for people who never got to use it so we can compare the new Piapro to Voidol and VocaListener?)

Sorry for rambling, I don't know if other people will understand what I'm trying to talk about.

 

[mobius017]

But I don't really understand how using Cherry Pie (the vocoder) would be linked to the actual tuning part in the new Piapro, it's most likely going to be like an updated version of VocaListener where it detects the start/end of words and you can enter phonemes over the blank music notes, but I don't know. (Especially since AIST, who made VocaListener, is now working with Crypton.) (Maybe I should make a post about how VocaListener works for people who never got to use it so we can compare the new Piapro to Voidol and VocaListener?)

Yeah, that's a point that confuses me, too. The impression I've gotten in the past has been that Cherry would be basically like a voice changer--you would speak into it, and the output would basically transform your voice into that of the character you picked. I don't remember that we've seen anything about how it integrates with music production, although they list it right there along with Vocal Drive like it's part of one cohesive package.

Since we have voice-recognition for things like phones, Alexa, Siri, etc., I don't see why it couldn't theoretically pick out phonemes from a microphone or recording (so it might be something like "get recording" - "do speech-to-text lookup" - "convert to phonemes" - "render as chosen synth"), but it seems like it would increase the latency somewhat. Plus the issue with actually getting it to recognize the spoken words accurately and tune them correctly. Neither of those things are good for real-time vtubers.

Given that, I've always sort of thought that Cherry might have two modes--one for vtubers (real-time conversion), and one for producers (voice input for lyrics and semi-automatic tuning that could be refined in the editor afterwards).

Assuming, of course, that Cherry is really intended to have a role in music production. The other option is that it's purely for vtubers. Which is kind of convincing to me, too, since in the past it's seemed like it has capabilities that include the ones offered in other parts of the package (Doesn't it have Vocal Drive-like capabilities itself or else have somewhere where it can call on Vocal Drive? I thought I remembered that from a screenshot somewhere. If that's the case, it doesn't seem like a plugin to use in music alongside Vocal Drive, it's more like a standalone product that borrows from the abilities of a "lesser" plugin when needed.).

 

[Nezuh]

Just had the idea could one use cherry pie on vocaloid or piapro to get English vocaloids closer to the Japanese version I'm thinking Len English

You don't need Cherry Pie for that. You need magic.

 

[Prism]

Do we know if cherry pie does actually vocal synthesis of is it just a filter. It kinda reminds me of antares throat ( Throat EVO - Alter Voices with a Physical Model of the Human Vocal Tract ) but with vocal presets for different characters

 

[uncreepy]

Do we know if cherry pie does actually vocal synthesis of is it just a filter. It kinda reminds me of antares throat ( Throat EVO - Alter Voices with a Physical Model of the Human Vocal Tract ) but with vocal presets for different characters

They've been calling it "vocal analysis synthesis". So the analysis part = like VocaListener/Beef Jerky/Songle (it detects where notes start/end and the pitch). Synthesis = produces the character voice.

Throat Evo seems like it gives you a picture of the human throat where you can stretch notes and make them go up/down in pitch (I guess it's Antares' equivalent to Vocaloid style tuning).

I think you might be on to something! When nyanyannya posted their two demos of Kaito with Cherry Pie, they posted a tuned and untuned version. But we were scratching our heads wondering how you tune in it. Based on the demo, Cherry Pie can be used like a normal vocoder (voice changer), but nyanyannya mentioning tuning confirmed you could somehow tune the results. If it's similar to Antares Throat EVO, that would explain how Cherry Pie/the new Piapro will work.

I tried to look it up, but I couldn't find out what type of synthesis Antares is using (ex: HMM).

 

[Prism]

Antares Throats really doesn't use synthesis it uses a combination of noise generation (breathyness) and pitch shifting with a forment shifter with a envelope filter

 

[Prism]

Sorry for the double posting but have we heard anything from the new crypton Synthesizer like a vocal demo

 

[uncreepy]

Sorry for the double posting but have we heard anything from the new crypton Synthesizer like a vocal demo

All we got was one demo in March (which I've linked to like 20 million times in this thread) that showed Cherry Pie and Vocal Drive. The only other clips we got were collabs last year Miku did for like Shinkalion, "NEWDAYS", the Tokyo 100 Year thing. But we haven't gotten a second official demo. Even at the presentations Crypton did at synthesizer/tech demonstrations, they only talked about the demo from March and had posters of it.

They said there will be news at the "end of the year" (maybe at Miku Symphony?), so just sit back, relax, and wait for the news. They've just been really secretive about everything. They're not gonna sell it til next year, either, so the announcement might not contain satisfying information.

 

[Megami Chan]

Wat's been making a lot of new tweets!
A lot of them seem to be more based around costs (mostly around models it looks like?).
I don't really trust myself to translate them all perfectly, but they seem at least kind of useful(?)
Maybe more important towards Project Sekai but I'm not sure if they all have to do with that.
Does anyone skilled with translation think any of them are important?



Also, I wasn't here to see if we had gotten all of these models yet (when I was last here, the MEIKO one was still underway), so here they are together.

 

[Ceres]

Kaito is just... T-posing back there, I'm sorry but I can't help to find that hilarious

Anyway, slowly but surely they are all finally here, I do wonder if their usage will be expanded outside of Miku Symphony and other small videos since Wat has put a lot of focus on finishing them (Although am I the only one that finds the area where Meiko's braid should be a little weird?) who knows if they become concert models...

 

[uncreepy]

ALL 3D MODELS COMPLETE:

All the members were completed, on time.... They're low cost and are... even expected to be related to a more detailed matter!

It's nice to finally have everyone finished. They've had the Miku/Rin/Len models done since like... two Miku Symphonys ago. But why is everyone posed except Kaito? He's in a T-pose and they tried to hide his arms behind everyone, that makes me annoyed!

============================================================

LOW COSTS 1:

...This is mainly for the reason for the basis of low budget matters, but,, from now on, I rather want to make it a rule to try our best at going to even incorporate various [3D] models!

============================================================

LOW COSTS 2:

A matter that goes back 3 years...

Company president: "What? Why are you making CG models, Sasaki-kun?"
↓
Me "Ah, it's necessary for having them talk at the vocal synthesis technology demo~. The span of this matter will broaden!"
↓
Company president: "Ahh, that kind of matter, huh"
↓
Me: "Including the voice demo, we're managing to get by with low costs!"

...it was that kind of sequence of events

============================================================

MISSED OPPORTUNITIES DUE TO THE COST OF INCLUDING MANY CRYPTONLOID MEMBERS

(Or should I say a long time ago the cost was too high > <..., even if the discussion became wanting to use all [Crypton] characters at the beginning of the plan, there were many times 3D CG promotion opportunities were missed due to the problem of work cost.... and of course there were problems with the cost of drawing all [Crypton] members.... including matters that even include Project Sekai, the plans that are going to include many characters will increase)

Maybe they haven't been ignoring other Cryptonloids because they are biased towards Miku, it's just that it cost too much to include everyone, so they had to decide how to focus spending money (and it ended up having to go toward Miku). I think this would be a reason for the real-time 3D so they don't have to pay Sega.

============================================================

MORE LIMELIGHT FOR ALL CRYPTONLOIDS + NEW CHARACTER IN THE NEAR FUTURE?:
When I click the tweet, it says it's unavailable, so I'm putting a screen shot here to preserve it. (The preview of it on the timeline works, but not when you try to click it?)
Edit: I guess it's just me having this problem with Firefox. I can see the tweet in IE and everyone in the Discord can see it.

For the many number of times I'm involved, "Hatsune Miku" is by far a clear lead... but what I am thinking I want to expand on is, even Miku and even Rin/Len and even MEIKO and even KAITO, all members! (In the not-so distant future... and maybe make a new character)

============================================================
EXTRA INFO ABOUT THE 3D MODELS + PROMOTION:

*Addition: The CG model maker is, "tumidango (つみだんご)". We are receiving the help of many other people / We won't use other MMD models. Instead we will properly think of balance / These CG [models] are going to be used to endorse other characters even if the project proposal & budget is for "Miku promotion only"

Tumidango's Twitter: https://twitter.com/tumidango

 

[mobius017]

Maybe they haven't been ignoring other Cryptonloids because they are biased towards Miku, it's just that it cost too much to include everyone, so they had to decide how to focus spending money (and it ended up having to go toward Miku). I think this would be a reason for the real-time 3D so they don't have to pay Sega.

I'd guess the production costs might be associated with a few of the things we wish would be improved about concerts. It could factor into why the inclusion of new songs, or songs by Luka/Rin/Len/Meiko/Kaito, is slower--the additional expense of getting the rights to the songs, the animation, whatever else. Miku's popularity is definitely a factor, too, but I think those things help explain why it seems like there's usually only 1 new song a year for any of the CFM 'loids (the theme song) at Mirai. The pool has expanded over the years, but lots of the same songs from prior years get re-used.

MORE LIMELIGHT FOR ALL CRYPTONLOIDS + NEW CHARACTER IN THE NEAR FUTURE?

Yay!

 

[uncreepy]

Amano gave some info that the "new character" in the near future might be for Project Sekai and not necessarily a new Vocaloid.


Edit: We'll have to wait and see. I thought maybe the girl on the promo art would be "you" as the player (or maybe there is a girl and boy player you pick from). But if it's related to the Cryptonloid world, hmm... I dunno what to think. I'm excited either way.

 

[First Sound Future]

Since Wat said there will be a new character in the near future and we will get more news regarding Project Sekai in late October (iirc?), maybe it's possible this new girl will be revealed then or sooner?

Having an actual human character that's not a vocaloid be somehow related to the Cryptonloids will be interesting but I'm thinking it will probably be the "MC" of the game, like a player character. Sega's latest game on smartphones had the option for you to choose a male or female avatar, maybe something similar will be available in Project Sekai?