No.1040
>>1038did you add an "a" sound before the breath? utau can't generate proper frq files if there aren't detectable pitch frequencies in the sample.
>>1039i don't have any demos that i can share but i'm pretty good with otos; maybe i can take a look when everything's done.
No.1041
>>1040the UST isnt mine but afaik there was no 'a' sound. if you mean if i added an 'a' to the alias, i didn't.
you can actually hear the breath working well in the first sample i sent
>>1016 but it no longer works so i probably fucked something up.
No.1058
We're getting closer and closer to this thing's final release, personally i wouldnt add anything else to the design its fine as it is. Godspeed anons.
No.1066
i think most people here will be opposed to it but i think the design should be made to look a bit more tomboyish. the current one is good but it doesn't fit her androgynous voice.
No.1067
>>1066Insanely rough sketch and i suck at anatomy but i came up with this. The head is the only thing that really changes, i kept the same outfit.
This way you can't tell if it's a shota or a girl with short hair.
No.1068
>>1067also I didn't want to make her look too much like Flower (she already sounds a bit like Flower!) so i didn't give her the same type of streaked hair, instead I gave her a single orange highlight similar to what Ryuuko from Kill la Kill has, picrel.
I apologize in advance to the original drawanon if my design ends up becoming the final one.
No.1077
>>1067Based on this and the anon mentioning they'd prefer a more "tomboy" design I toyed with various ideas, while I still personally prefer the current design I think it would be cool if we had one "long" strand of hair akin to Asa Shigure (who is also green haired oops), but I also just shoved a whole bunch of boys with interesting hair styles that have some "longer" hair features if people wanna toy with that.
I also added a star for if we go with the name Sena, since the connection with stars is also appealing to have in the design if we keep that.
Obviously since I have no clue at all how to draw I just shoddily added some messy lines.
No.1081
>>1077I also prefer the current design but i don't think it fits the voice. The voice is more boyish than the design. I think another thing we could do is use the current design but make her look more like an adult, like Fuiro who also has a deep voice.
No.1099
>>1077I think the star is too much, it makes her look too orange, but i like the hair idea
No.1128
The final version will use drawanon's design.
As far as i know these are the only missing phonemes so the voicebanks is almost 100% done.
>ra, re, ri, ro, ru
>pya, pyi, pyu, pye, pyo
>mya, myi, myu, mye, myo
>rya, ryi, ryu, rye, ryo
No.1131
>>1130KAWAIII
No.1132
>>1131Thanks anon. I don't intend on this becoming the final design (unless people really like it that much?) but if she ever receives an update like CVVC, VCV, Diffsinger even it could come in handy.
No.1136
>>1130I just noticed that I misspelled her name
No.1157
>>468Can i try my hand at a redesign?
No.1206
>>896>>901>>912I actually believe we can make something like this work in the current year and i have some suggestions.
Obviously faking a SynthV can be hard but not impossible. The only way we can currently make a fake SV is with RVC… an 'RVC'loid, if you will. It'd basically just be a RVC bank with a character design attached to it. RVC when given high quality training data can sound convincingly realistic.
>But DT bans SV as-Shh… listen, we don't have to use SV as the reference audio. Again, what's the primary selling point of SVS2? Realism. We can just use a human singing and tune the input/output with something like Melodyne or VocalShifter. The design for the RVCloid can be literally anything. As far as I'm concerned they don't have a bias towards Asian artists or anything, SOLARIA doesn't look like she was illustrated by an Asian artist (I could be wrong, but the overall style is pretty Western looking).
No.1207
>>1206Another thing we could do would be to fake a CEVIO. Now here, we would actually need an illustrator who can mimic anime-like art styles because I can't think of any CEVIO banks which use Western artwork. Alternatively, commissioning a Japanese artist is an option but with how broke 39chanons have proven to be that's unlikely to happen. With CEVIO we would just have to make a DiffSinger or NNSVS/ENUNU. How we would get people to take the bait is beyond me since the news source would probably have to be some Japanese Twitter account. But DiffSinger and NNSVS/ENUNU's engine noise play in our favor here. CEVIO is VERY noisy so it should not be hard to fake.
Here's how both sound for comparison:
https://www.youtube.com/watch?v=TW9hOIwXGrUFaking a V6 would be a bit of the same, except it's not very convenient to fake a V6 for the fact that:
>1. No one cares about V6>2. V6 sounds flat as hell, can you even achieve a sound like that with OpenUTAU?>3. No one cares about V6>4. NO ONE FUCKING CARES ABOUT V6 RAAAHHHtl;dr: we can pull it off
No.1225
>>1206I also saw this really good fanmade Neru voice today! From what I see they used Vocoflex + morphing Miyamai Moca with samples from a very old vintage singing synth, since we are using human input (Sena's VP, unless you'd like to use an UTAU render instead) it should be way less noisy and similar to an AI vsynth as well. Along with not breaking DT's terms because it uses a VST instead of an SVC.
https://x.com/leastbehuman/status/1883410163354538413I also really love the idea of faking an SV instead of a Vocaloid or VIPPERloid because the "fanmade VIPPER" and "UTAU inspired by old weeby amateur UTAU from the 2000s-2010s" UTAU thing is kind of played out, hell I was even the voice of my friend's own UTAU with a similar concept. I agree Solaria has a completely feel to the average vsynth with a very Western looking design (in a good way though) and I'd love to do something that's more of an original and Western take on fakeloids. Ideally she'd have an art style more like one of the cuter League of Legends or Overwatch characters, or Sakimi-chan/Ilya Kuvshinov/Skullgirls/90s and 2000s cartoons but yes I am also broke and lack commission money for that lol
More ideas in picrel, it was too long to post
No.1227
can we give eiiru h8rs their own thread? I'm here for her
No.1228
>>1227They are as non existent as her voicebank
No.1229
>>1225I meant a separate voicebank not Sena. Sena should be completed as an UTAU and be its own thing, then we can finally pull the fake SV prank. I feel like it would probably be easier to fake a CEVIO one though as i said in the original post, it's more noisy than SV.
No.1239
>>1229Ohh alright my bad. Should we make a separate thread for that? I don’t want to derail.
Also, since the western fandom hates lolis, can we make her a MILF? No.1241
>>1239Yeah, would be nice to have a different thread
At least give her massive boobs No.1544
otoanon here. this is what's missing:
>myi, myu, mye
>rya, ryi, ryu, rye, ryo
after they're all recorded i'll post the voicebank here but for now i gotta wait on vpanon.
No.1573
is this ever going to be finished
No.1607
After many months i am proud to announce that Sena's CV voicebank is finally finished. Thank you all for being patient.
DL:
https://files.catbox.moe/zs49pu.zipDo note the following:
- This is my first time otoing so the .oto is a bit messy. Hiragana characters don't work so you'll have to either convert your USTs to Romaji or fix the oto yourself to support Hiragana.
- The breath sample as mentioned previously is fucked, at least with moresampler. I have no idea how to fix it.
Other than that it should work just fine with any song.
No.1608
>>1607HOLY SHIT SENA IS FINISHED?
No.1611
>>391 >>1608Sena released before GTA 6 !!!!!
No.1612
>>1611Sena released before SatapanP's new song
No.1624
>>1607aww we didn't do the good name :(
No.1629
>>1627yess, mao-tan supremacy. it's just like a JP fanloid name
No.1655
Someone please update her to support Hiragana
No.1659
>>1618Small mistake, line 34 of the oto.ini, 'gyu.wav' is aliased as 'gyo'.
No.1660
i'm too tired to add hiragana support but i fixed this:
>>1659https://files.catbox.moe/esirhk.ini (rename to oto.ini)
No.1661
I want someone to draw Sena eating Surstromming while the other 'loids/utaus are like "how the hell are you eating that garbage?"
No.1694
Alright this
should be the final version of Sena.
- Added Hiragana support
- Made the transparent image bigger + removed the "77 chanloid" from Sena's arm since it's pretty irrelevant
- Removed the leftover Hz conversion batch file included in the previous zip
The breath sample is still kind of a mess but if it's short it sounds fine.
https://files.catbox.moe/t77r0x.zip No.1735
>>1694- Fixed the "shi" sample being very stretched
- Minor changes to Sena Transp.png + included original image
https://files.catbox.moe/cc9xdd.zip No.1738
anons, what should sena's likes/dislikes be? the bio was written a while back but not the likes/dislikes.
No.1739
>>1738Likes: Earth
Dislikes: Earth
No.1740
>>1739I agree but can we also have her like cats
No.1741
>>1738likes: opening doors, the fitnessgram pacer test, french fries
dislikes: butts, thesaurus
No.2056
>>2055I feel like she needs better clothes but other than that it's cute
