Let's fake a new voicebankMikudayoSat, Feb 1, 2025, 03:39 PMNo.1270
Alright, /leek/ has proven that they're capable of making and not finishing an UTAU, but due to popular demand let's have another thread for the "fakeloid" project mentioned in that very same thread.
As we already know, it's impossible for filthy 39chan gaijin to fake a VOCALOID the same way VIPPERs did Teto, Tei, Ruko and the rest. But it's entirely possible to fake a voicebank for another engine like SV, V6, Cevio and Voisona (though the latter two would likely not gain much notoriety if carried out by westerners).
One of the ideas raised in the original thread was to create an RVC or Diffsinger based on an anon's voice.
I genuinely believe this might not be that hard to pull off. 39ch isn't a huge site so it's not like many people would snitch on us and making an AI voicebank these days is piss easy.
I feel like this is way too elaborate of a plan, especially for /leek/ standards but hear me out. >An anon creates a Vocaloid news account on Twitter in the same vein as PositiveVSynth >The account is created BEFORE we even consider the fakeloid's details or design. Anon just keeps the gimmick going for as long as possible. >The voicebank is created. For this an anon would have to record samples of themselves singing in Japanese and English, in different styles, it would also have to be approximately 3 of each (3 Japanese soft audios, 3 English soft audios, 3 Japanese power audios, 3 English power audios, etc) you get the idea. >After the voicebank is created (diffsinger/enunu/rvc) another anon will create the design. This should be after the voicebank is finished to prevent what happened with Sena. >An anon purchases a website/domain for what would be an indie studio (it could also just be a wordpress site with a domain) and creates a blogpost detailing how they made a new voicebank for SV >Anon's vsynth twitter account reports on it >??? >Profit
Speaking of Solaria, i think the fakeloid's art style doesn't matter too much so long as the final version doesn't look super amateurish. The majority of SVs are drawn by asian artists but if we commission literally anyone who can draw well it should not matter.
One thing we could do for the design is use a character creator as a base of sorts. Momo and Defoko's original designs were created in Animegen (picrel) and they later received proper illustrations.
>diffsinger based on an anon's voice this sounds like the best option for a voice people might use + it prevents any rulebreaking shit. i can't help drawing or publicity wise but i could record the vb my voice isn't very kawaii desu sadly but i fit the mature sounding part a little… this is an rvc bank from a year ago
I think that while we should put in the effort we should also make it a bit clear that it's a joke character. All vipperloids have joke traits like Sukone Tei being a coomer, Teto being a chimera that sounds like Doraemon, Ritsu being a male with tits, etc.
How about making her an actual menhera? And I'm not talking yandere or any of those archetypes. Make her actually mentally ill, make her a schizophrenic who thinks she's a popular idol or something but is actually a literal who or something like that.
>>1290 This btw. We should use DiffSinger instead of RVC. RVC is too closely associated with the Drake and Kanye AI cover people. Besides, it would possibly be the first bank for that engine to become popular if we pull it off. That's what happened with Teto. She was so well made that she sort of became the poster child for UTAU. It would also be nice to see more DiffSinger originals in general cause i can't think of any producer who uses that.
i feel like you guys are thinking way too ahead. the vpanon from earlier hasn't replied nor sent any of the required audios for diffsinger. we should also let more people audition perhaps. i do really like the first anon's voice but idk about immediately choosing the first one when there could be more people willing to participare. not to mention 39's slowness can cause people to abandon the site and never come back like sena's artist.
the designs are also varied and cool but only the first one feels like it could be a synthv imo (if drawn by a japanese artist). keep in mind synthv designs are very simple for the most part (with tsuina probably being the most overdesigned) so while the yami kawaii aesthetic we're going for is neat we should not make the girl super complex to draw.
>>1310 This is true. Originally I intended to make multiple designs anyway, but my friends really liked it and encouraged me to post it here even though I agree mine is a bit too weeby for a SynthV lol. Also rushed a bit as I had somewhere to go. There’s no time limit right? Great if true, maybe we could add “aspiring kaigai (overseas) idol” to her concept/bio as well to explain why the designs are so Japanese-looking.
>>1310 We also need someone who can train the voicebank which is a whole rabbit hole i personally don't want to get into. RVC from experience isn't hard but DiffSinger on the other hand…
There's a few resources on the DiffSinger thread on this site if that's anything. But that thread's kinda dead so good luck askung for help.
As much as i'd like a yami kawaii utau i think she should look more normal if we're making her canonically schizo. It would make her more realistic, like Scottei who's literally just a girl with white hair. Synthvs also look normal in general as mentioned above.
>>1319 Could be, yes, but there isn't such thing as a schizophrenic fashion style. Picrel was a schizo and if it wasn't a known fact I wouldn't have guessed by just looking at him.
Did the vpanon from earlier vanish? Hope they weren't put off by our proposal for a schizoloid, having them be at leastly slightly offensive like the original VIPPERloids is part of the fun.
>>1338 sorry i haven't disappeared! i just thought i'd wait before doing anything in case anybody else wanted to be the vp. idm vipper-like ideas, though i have an english accent so i'm not sure how well that would fit the design. might be overthinking but just putting it out there in case there's any issue. i'll get the recordings done asap. i'm no singer but the diffsinger should still sound alright. also just as a disclaimer i'll delete the post with the link to the recs after the DS is finished, it won't mean i pulled from the project
>>1274 about this, i was wondering if we could message positivevsynth about this? there's a chance they'd be a bitch and ruin the prank but i figure they know about teto's history. they might find it fun to take part in something like that
>>1341 oh i meant i have a uk english accent lol, you're still right though
>>1345 i doubt pvs is a bad person but i also think messaging any existing account is a bad idea lol. better to make one from scratch.
>>
SV News !!Eek6C43g5MFri, Feb 7, 2025, 04:48 PMNo.1348
I'm currently not home so i can't do this now but im up for the task of running the SV news account. However I'll keep it a secret. If people outside 39chan find it and happen to follow ;) it, good, bait taken. If people on 39chan happen to find it. That's cool too, just don't blow my cover if you somehow connect the dots and start to figure it out.
>>1339 >>1345 Ganbatte britanon. A British SV would not be a terrible idea considering all of the existing ones are very American. The Vocaloid era had Sonika and Oliver but SV has nothing afaik.
>>1373 I have $200 in savings at the moment and will get more in the future, it's for emergencies but since I'm a huge vsynth autist I'm willing to help out (esp. if said commission ends up being around only $50-100)
I agree with the anon who brought up Jirai Kei fashion, a lot of people like this style so i feel like having a design based off it would draw some attention to the voicebank, it also kinda fits vpanon's voice in my opinion. I think she should look more like picrel. Maybe with twin drills (not like Teto's) or twintails.
I also don't know if we should strictly make her a milf with big tits. I know the argument is that the western fandom hates loli but flat doesn't mean loli.
To the few people here (if any) who know how to make DiffSingers - Are multilanguage voicebanks possible or would we have to make two separate voicebanks one in EN and the other in JP?
If our plan was carried out successfully, would this help popularize DiffSinger as an engine?
A decently sized chunk of the vocaloid community already knows about its existence but very few voicebanks exist due to how hard it is to train a model, as for original songs, there is very few as well due to how Cevio and SV are more attractive options for producers.
>>1286 If you think about it, DiffSinger's imperfections do not really matter in the end. We don't have to create a fake demo. Have you seen how Uni SVS2 was announced recently?. Just a single tweet in Korean with a release date and a link to Characterplanet. Our Menhera-chan can be announced similarly, except first we need two twitter accounts: The first one would be for a fake company, which is yet to be named. I would suggest "Unstable Sounds" in reference to our character's poor mental health. This is where the voicebank would first be announced. The second would have to be for SV/VS news. Have you noticed how PositiveVSynth often QRT's anouncement posts? That's what we should do.
So the plan is simplified even further by this. It's all a matter of not fucking up.
First of all: the two Twitter accounts need to exist first, if you have any unused accounts made pre-COVID go empty them and use those, it will be more convincing. After that /leek/ has to finish the voicebank (that's up to both VPanon and 4090 guy) and then decide on a character design, there's already plenty ITT. When vb/design are finished the Twitter account can finally announce the voicebank and maybe link to a website that tells you "lolz u fell for it" and has a download link for the DiffSinger voicebank. The SV news account would retweet this and we're pretty much done.
>>1472 One major problem, SV News already exists. (https://x.com/SynthVNews), there being two SynthV news accounts will most likely raise suspicion.
Some possible solutions >Run it in a different language so there's an excuse for two SV News accounts existing, I might know some Spanish speakers. >The pre-existing SV News account does exclusive interviews with VSynth companies (https://x.com/SynthVNews/status/1756735055399907443 [Embed]
), we could do an interview with them instead of running our own news account, we just have to decide if we're going to be honest to them about it being a prank or LARP as an actual company.
>>1493 >We could do an interview with them instead of running our own news account Why do people in this thread actually think this is a good idea? Our cover will be blown immediately. People on Twitter hate fun.
>>1494 I agree with you actually, I just can't think of many alternatives. How do you feel about my other idea? It could be something like "@SynthVNewsES."
>>1495 If it was English or Japanese then perhaps but I feel like any other language would raise suspicion. If our voicebank was Spanish then maybe that would make a bit of sense but it's EN/JP.
>>1496 Idk, there's that one upcoming Spanish Opera vocal from Eclipsed Sounds + SV has had Spanish support for a while (I'm not big on SynthV though, so correct me if I'm wrong), along with me seeing tons of Spanish speaking SV fans on vocatwt.
>>1566 i'm vpanon, I'm still here! the softer and normal jp + en recordings are done, i'm just trying to figure out the best way to do power since i usually can't sing too high or hard. i've also been waiting because of the talk about other languages in case i need to record anything else
Do we even have a name for her? I like the idea of Lana being her actual name but Himari being her stage name. (If we’re going with the kaigai idol route) I like the juxtaposition between a dark and gloomy design with a name that means “ball of light”
>>1755 We've been calling menhera-chan but she doesn't have a name. I don't know how to feel about Lana but i do think an American first name + Japanese last name would be cool or a fully American name (preferably something that doesn't sound silly like Ashley Taylor from Magia Record)
>>1756 * reminder that her name doesn't even need to be a standard human name cause what the fuck is a GALENAIA or an ASTERIAN
though i feel like giving her a name like that would make people suspicious that we're ripping off Eclipsed Sounds. "Unstable Sounds" already sounds a bit similar.
>>1757 other companies also kind of do it. 'ANRI', 'Stardust', 'Minus' and all that. i think we could get away with it if we don't make it entirely obvious
Hmmm if we’re gonna do American first name + Japanese last name I think it would fit better if the first name already sounds sorta Japanese ( Hannah, Emi or Mari for example)
>>1763 I like the jirai one but i might be biased because i love that fashion style. But overall i think its the most realistic one if it was drawn by a jp artist.
before this thread moves on i wanna put in my two cents this is uncoloured for the time being, but i wanted to try make a more complicated design since that seems to be really in at the moment (vtubers whatever) a bandage motif i thought would be a good idea too, nailing in something like that would allow recognisability. i'm gonna try colour, but feel free to try colour her yourselves, or do literally anything.
>>1765 small criticism but i think you should get rid of the eye bandage but keep the leg ones. eye bandage kinda feels like its too much, especially when the hair already covers that eye.
>>1772 >>1770 thanks for the input! the face looks better without the bandage i think, and i tried to make the hair streaks a bit more pink (i would of done more but it made it stand out too much) thoughts so far?
Considering our menhera girl is canonically schizophrenic it would be fun to play with the idea of a wholoid. Menhera-chan could have delusional thoughts about being popular, to the point where she thinks people stalk her due to how popular she is, while in reality no one knows her.
>>1761 Personally i think we dont need an american or british name at all. She can be Japanese. Kasane Teto is Japanese yet she can sing in English too.
sankyu all for being so kind :) can't wait for her to get fancy artwork >>1807 yeah (didn't have that in mind) (funny as hell though so i'm pretending i did)
>>1815 Cute, but tbh with all these designs, no one being able to decide on a specific one and vpanon showing no signs of activity I'm beginning to worry.
>>1810 How accurate is this name though? Because I remember before Sena Tsukuyomi was chosen for the last UTAU, Hoshi Utara was one of the suggestions and it sounded "too foreigner".
>>1852 Also i feel like something that roughly translated to "Prisoner of the mind" or similar things would probably make more sense since menhera-chan is canonically a schizo.
some things might need fixes. i would of added the headpiece, but it cluttered too much with the hair so i decided to get rid of it. sorry fellow drawfag but let me know how you guys feel about this. tits too big? too small? anything goes
Out of curiosity has anything like this ever been attempted before by the western Vocaloid community? I know VIP@2ch did this exact thing with basically every VIPPERLOID but has 4chan ever tried to prank the community with a fake 'loid? I'd be shocked if not.
SV News !!Eek6C43g5MWed, Feb 26, 2025, 08:21 PMNo.1885
Okay anons i have a masterplan so hear me out. While we're still in the process of making Menhera-chan (vpanon and gpu guy gone MIA, character render still doesn't exist, etc) I've made an account on Twitter for a fake VST company specialized in audio stuff. I've showed it to a few musician friends and they agree that it's realistic enough to fool them. >What about Unstable Sounds? I happen to know the anon who runs the U_S account and let them know about my plan. We can't do Unstable Sounds for a couple reasons: 1. It sounds too much like Eclipsed Sounds, and as a Graphic Designer, the logo also isn't very convincing (sorry). 2. I have been a producer since 2016 and follow VST news and all that good stuff. I know what a VST company tweets like. I'm not entirely sure if the anon behind the U_S account does, though if they do that wouldn't surprise me since they are also a musician themselves. 3. The golden rule here is to maintain anonimity. The fact that we linked to the U_S Twitter account already gave us away so any lurker here can just go "oh that's just a fake 39chan account". tl;dr: I am in charge of managing the fake company account and have lots of time to post about random VST shit while you guys work on menhera-chan. If i'm lucky i might also have a friend set up a troll website for it. >What will happen to the SV News account you teased? The anon who runs U_S will delete U_S and create an account for Vocaloid news similar to Positivevsynth. Since Ia'm bilingual i will also manage a similar account but in another language. When menhera-chan is finished both accounts would report her announcement/release. >Isn't this way too elaborate for a troll vocaloid? Yeup….
>>1885 who cares if it's too elaborate it sounds funny as hell let's do it
>>
SV News !!Eek6C43g5MWed, Feb 26, 2025, 08:43 PMNo.1887
>>1886 I think you guys should focus on finishing both the artwork and voicebank for now. Look into Diffsinger tutorials and all that crap. Join discord servers to learn if necessary but DO NOT mention the voicebank, 39chan or any of that there. Don't blow our cover.
I'll handle half of the twitter/website related stuff, while the U_S anon will handle the other half.
>>
SV News !!Eek6C43g5MWed, Feb 26, 2025, 08:52 PMNo.1888
>>1862 Also to whoever ends up commissioning the artwork for Menhera-chan, go with this design. You have multiple anons agreeing that this should be the final design even before this one was posted. Besides it's realistic enough to pass for a SV.
Go on Skeb or commission some foreign artist who can imitate the standard "Danbooru frontpage art style".
"____ is a 23-year-old idol with schizophrenia who lives her life in a blend of delusion and dreams. She firmly believes she is a superstar, constantly surrounded by adoring fans and even a few stalkers, but in reality, she remains a complete unknown. Her obsession with fame is reflected in everything she does: from the way she dresses, to the constant (and sometimes incomprehensible) social media posts, and her belief that every passerby is a potential admirer.
Off-stage, she is extremely introverted and often locks herself in her room, shutting the windows and plunging the space into total darkness. She’s convinced that birds are actually government-operated drones, spying on her every move, and she believes that by keeping the windows sealed tight, she can avoid their surveillance."
>>1897 This is good enough imo. SynthV characters typically have heights and weights and birth dates right?
I want to suggest April 1st, 2001 as her birth date (or September 11th if anons are willing to go that far). As for the height/weight, i'll leave it up to the other anons.
>>1898 god if i had the skills to draw that sort of thing i'd be all over /otn/ believe me >>1905 the second one would be SO funny. i actually really want to go for that.
likes: playing tea time with plushes, ned flanders, makeup, balsamic vinegar, catnip, walmart, puzzles dislikes: birds, VOCALOID3, anything related to spying, wearing underwear
menhera-chan plays teatime with plushes when locked on her room. she ships herself with ned flanders and believes he is her greatest fan. menhera is also a big fan of walmart. she believes underwear is a lie fabricated by the government.
other of her hobbies includes resolving puzzles and planting catnip. she loves makeup and drinking balsamic vinegar. she dislikes VOCALOID3, since she believes that she's more popular than those "bastards".
>>1920 I like this idea but are likes/dislikes even a thing anymore outside of utauloids? I've only seen regular short bios, height, weight and gender.
Name: ??? Bio: >>1897 Birth Date: September 11, 2001 Likes: Playing tea time with plushes, ned flanders, makeup, balsamic vinegar, catnip, walmart, puzzles Dislikes: Birds, VOCALOID3, anything related to spying, wearing underwear
I think once we have the name, height and weight drawanon should make some sort of reference sheet and include that there and then the anon that has the money, if they're still here, should commission a JP artist.
(anon if you're still here commission a good artist dammit)
>>1983 samefag, I guess my best bet is Nono Toriko (脳乃とりこ?), it literally translates to >>1853 and it rolls off the tongue Toriko is a shonen manga and the word "adopted" before it's an actual name but that's probably fine
it's march and vp still hasn't returned. i hope they haven't quit the project entirely since i liked their voice. >>1975 i think its a bit over the top, you should keep the tea thing and replace flanders with some synth that isn't len because at that point you're just making tei. the rest can also go.
>>2066 >(minus the illustration side) Hello, I'm $200anon and I'm still here lol!! Ideally I'd like to wait til next month if possible but if you need the art *ASAP* I'll try and make it work. I was waiting for the voice to be finished
>>2073 From the guide you posted: <Please don’t autotune/Melodyne your recordings either, it just makes them worse for AI by introducing pitch-shifting engine noise, and we want to minimise any artefacts in the training data. If you intend to train an autopitch model, know that Diffsinger compensates for the off-pitchess of your singing automatically either way, which further defeats the need of tuning the recordings beforehand.
>>2073 My question is whether these should be separate voicebanks (TORIKO EN, TORIKO JP) because afaik you can't have two languages in one voicebank with DiffSinger, just different expressions (Normal and Soft in this case)
Here's a tutorial video which shows this (in Spanish because i guess the English DiffSinger community doesn't exist) https://youtube.com/watch?v=TKZvOLhBQWM [Embed]
Maybe we should wait for the next SV announcement/release to announce TORIKO. A lot of the recent voicebanks have been announced on the same day so it would make the prank more believable (and it would also give me time to post more news).
>>2064 Name: TORIKO Bio: <TORIKO is a 23-year-old idol with schizophrenia who lives her life in a blend of delusion and dreams. She firmly believes she is a superstar, constantly surrounded by adoring fans and even a few stalkers, but in reality, she remains a complete unknown. Her obsession with fame is reflected in everything she does: from the way she dresses, to the constant (and sometimes incomprehensible) social media posts, and her belief that every passerby is a potential admirer.
<Off-stage, she is extremely introverted and often locks herself in her room, shutting the windows and plunging the space into total darkness. She’s convinced that birds are actually government-operated drones, spying on her every move, and she believes that by keeping the windows sealed tight, she can avoid their surveillance. Birth Date: September 11, 2001 Height: 145cm Weight: 48kg Likes: Playing tea time with plushes, Public urination, Tile patterns (Fetish), LSD Dislikes: Birds, Swimming Pools, Wearing underwear
>>
SV News !!Eek6C43g5MMon, Mar 3, 2025, 03:08 PMNo.2105
I have good and bad news regarding TORIKO. Let's start with the good news: first of all, the aforementioned accounts are doing WAY too well. Better than i expected, in fact. So if i post TORIKO's announcement people will see it. The other good thing is that we finally found someone to draw TORIKO's final artwork, but you won't get to see it until the voicebank is finished and the announcement post is up.
The bad news is that this is going way too well. This would be good, but keep in mind that the Western fandom isn't like the Japanese fandom. Making an offensive UTAU and passing it off as an official SynthV is like throwing yourself into a tank full of sharks. The idea is to have a website with the information from the thread (see >>2092) so people know that it's just a meme and nothing more but again, this is the Western fandom we're dealing with.
I'm part of the group that's taking care of promotion, art, etc. The website for Toriko has been purchased, it just has to be finished now, and as mentioned above if its done before April then amazing. If not the prank will still be carried out.
Cheers and good luck to the group that's working on thr actual voicebank.
To recap >Character details Finished >Design and final render Finished >Voicebank Recorded, but not finished >Social Media Accounts Finished(?) >Website In progress
is that correct?
>>
SV News !!Eek6C43g5MMon, Mar 3, 2025, 04:58 PMNo.2110
>>2109 Yeah, this is correct. A lot of the more technical stuff is being done behind the scenes by our team for organization reasons and to prevent snitches, bad actors, etc but don't worry at this point there's no way the plan will fail. Unless it backfires as mentioned here: >>2105
The promotional material for the Twitter accounts is also finished too so the moment the voicebank is trained and finished that'll be it, TORIKO will be public.
We're planning to have everything set up for April 1st but if that can't be done at least we tried.
In the worst case scenario that GPU anon fails his training or something you can attempt to commission someone who trains DiffSinger voicebanks as a last resort but chances are they'd fuck us over and cause drama so i don't recommend it.
>>2105 I was thinking maybe the meme information could be semi-hidden on the site? Like make the text the same color as the background (ex. white on white) Also, I’ve been diagnosed with schizophrenia and I still love her, so they can cry all they’d like. There have been way worse depictions of us (Movies that act like all schizos/bipolar etc. people are murderers..)
I can also assist with the site’s HTML and CSS if needed! (IDK JS or any actual programming languages right now though orz)
>>2140 If someone else does the labeling and configuration for training, I can run the training, but college is taking up too much of my time to be able to do the first two…
>>2145 probably just 5. everything other than the voicebank is finished and some anon can just watch a diffsinger tutorial / get a guide off some discord server (why the fuck is everything locked behind discord servers these days) and train the voicebank on their own.
vpanon here, i've pitched up the recordings so you can't tell it's me right off the bat lol. i thought about the whole 'being thrown to the sharks' thing and decided to have a bit of separation between the voice. honestly i assumed only a small amount of people would see this prank and laugh about it but >>2105 sounds ominous as hell haha
to me it still fits the design (and it might fit a troll vb to have a weird voice at least). she sounds like someone pretending to be cute as well, so it might fit her delusional idol thing a bit more? this is me coping about the decision but hopefully it's fine i thought i should send this now while progress on the bank itself is still very early
>>2155 Audio was very noisy and quiet so i did my best to clean it up. Its not perfect but it should be better, especially for whoever ends up labeling it.
>>2160 English version as promised earlier. Also quick note but i can't understand vpanon and i'm probably not alone so lyrics or song names would be appreciated. https://files.catbox.moe/bn0aph.zip
>>2161 the ones i recognise: world is mine = world kokoro mikumiku = miku miku ni shiteageru thousand = senbonzakura duvet = duvet by bôa regret = regret message (english cover) twins = servant of evil (english cover) the rest i'm not familiar with, some other anons might know
I have no idea how to train a DiffSinger (god bless the motherfucker ITT who ends up making it) but i made a mockup of what a finished TORIKO voicebank would sound like with RVC.
>>2175 >>2176 Perfect song for her. I was expecting her voice to sound more "mature" but i think i prefer this. The low quality of the samples also enhances it surprisingly enough.
I hope you guys find someone who can make the DiffSinger because this could be the definitive DiffSinger vocal just like how Teto is UTAU's. I dont know why you didnt from the start.
SV News !!Eek6C43g5MSat, Mar 8, 2025, 05:02 PMNo.2192
This is all i can reveal from the W.I.P TORIKO render, it's not 100% finished but it's getting there. The fake website is currently up although it currently uses placeholder images due to TORIKO's render being incomplete.
>>2183 As of right now the voicebank still doesn't exist and at this rate I'm thinking the possibility of it ever being made are lower than everybody was expecting. Our only option left is to commission someone to make it, which is a double edged sword because the Western UTAU community is infamous for being overdramatic and chaotic, and anyone could either A - snitch on us, B - decline our offer because "39chan is an evil 4chan ripoff for ugly fapper nerds who like kagaminecest diddy drake sex" or C - create the voicebank and claim that TORIKO is their creation and not ours.
>>
SV News !!Eek6C43g5MSat, Mar 8, 2025, 05:08 PMNo.2193
>>2192 tl;dr: train the voicebank yourself or we're fucked
Regardless of whether or not we have a voicebank by the end of March, the site will still go up and the purchase button will redirect to a rickroll. It's simple, but effective, and better than nothing.
Also twalam, I've seen your tweets. You voice a character. It's no different from being an actor in a film. People can dislike the character, but to harass the person voicing it is just silly. If they fuck with you, asses will be kicked.
SV News !!Eek6C43g5MSat, Mar 8, 2025, 05:27 PMNo.2197
>>2194 >toriko won't be made I never said this. I just said that the fate of her voicebank is uncertain. It most likely will be made but AFTER April, rather than in time for April. No one here seems to know how to train the DiffSinger VB so it's likely someone will end up commissioning it instead.
Some more TORIKO RVC testing w/ SV Teto input and Tsukuyomi Sena backing vocals. It has a lot of noise despite being trained on the cleaned up versions. I guess it's up to you to decide whether that's a good thing or not.
Also sorry, it's not the DiffSinger. I still have a bit of faith we'll get that eventually but not on time for April 1st. :(
>>2374 Get twalam a better mic. Set up a gofundme or something and i'll gladly donate. I don't care if RVC or Diff. I want to see this finished and with better quality than those furry diffsingers.
>>2375 Oh I'm not twalam I just made a private RVC using the cleaned up versions of the samples she sent, sorry for the confusion?. I also want to hear this voicebank in a better quality but beggars can't be choosers and keep in mind that Teto was very very low quality before her SV upgrade. Then again the SV was commercial, the UTAU wasn't.
>>2421 Alright, would you all ever want to approach a project like this in the future or no? God I wish I wasn't the only anon with money..it was either Toriko or groceries and urgent dental work..orz
At least we have Sena! Aren't there also tons of failed, unfinished, and voiceless VIPPERloids?
maybe one day we'll be able to do it fully, but even if we can't do that i'd be glad to just get the prank out anyway. maybe it'll attract someone who knows diffsinger and we can pick her back up
>>
SV News !!Eek6C43g5MWed, Mar 26, 2025, 11:35 PMNo.2455
I was gonna make it a surprise but honestly we're close to April 1st and I doubt anyone will come here to ruin the fun. Here's Toriko's final render, drawn by Melonenbrot. It was too good not to share.
The voicebank will sadly not be finished by April 1st, and at this rate I'm not sure if it will ever be made, but hey the website is up and works.
i'm thinking of alternatives that could be ready for the 1st and i do have an old utau i used the same pitching up thing on. this was intended to be slightly accented, though. hifisampler came out so maybe we could just do the prank with that? i don't know how it'd not immediately give it away though, unless the demos were on some part of the site that comes after they're called idiots/during. this is what it sounds like with hifi (https://vocaroo.com/16zrQ2kM2qRd [Embed]
>>2473 how about saving the same audios under different names, until they manage to form more than 50 minutes? that's the trick i use for RVC when i have no more audios available from a source, it could work here. sure, it's diffsinger and not RVC, but it could work.
Alright anons we couldn't get the voicebank done in time so this is the next best thing.
In about one or two hours the TORIKO announcement along with the website will go live on these two accounts and possibly more if the news spread. The non-english news account flopped so it's not worth posting there. What i need you to do is to keep an eye out and retweet, spread the news, give TORIKO the numbers.
April 1st is over but TORIKO's actual voicebank is being worked on. I got in contact with her voice provider, she's working on the actual DS voicebank so I guess at this point "VoxSoft" can be considered a team. I'll probably host the finished voicebank on the existing VoxSoft website once it's finished (assuming it'll be finished this year because the site will go down in 2026) then after that I guess someone could add it to the TCC page where Sena is currently being hosted.
Have you seen how Uni SVS2 was announced recently?. Just a single tweet in Korean with a release date and a link to Characterplanet. Our Menhera-chan can be announced similarly, except first we need two twitter accounts:
I also want to hear this voicebank in a better quality but beggars can't be choosers and keep in mind that Teto was very very low quality before her SV upgrade. Then again the SV was commercial, the UTAU wasn't.
