No.22290
It's easy to say we're in a second golden age of Vocaloid right now. Hundreds of songs are being made per day, small producers (including english and non-english ones) have been getting their moments of fame, old producers are returning to the scene, many beloved voicebanks are getting upgraded while 2ch VIPPERLOIDs are finally getting official voicebanks on commercial engines, if you can't afford VOCALOID there's hundred of synths available (Some are even free!). The fact that this site is so active in an era where imageboards are barely relevant anymore goes to show how many people care about Vocaloid.
But we gotta remember that just 10 years ago things seemed bleak. Producers were quitting, people were saying that Vocaloid was dying, many Vocaloid related websites were shutting down, and there was an overall feeling that Vocaloid was on its last legs.
Things don't last forever. As much as it sucks, this "2nd golden age" of Vocaloid will eventually come to an end and bring us back to where we were 10 years ago. So with that said, when we inevitably return to the "dark times", what do you think will happen and what could save the scene?.
No.22291
Better ways to do live performances probably. A lot of past live methods are more nichés and only involved Miku to my recall of ones i am aware of. Unfortunately instead in current trajectory might just get "AI" idols and music instead to eliminate producers. I dont think anyone will put up for that path though so it doesnt seem as bleak as it does at first.
No.22292
>>22291"AI producers" aren't really welcome in the scene so I don't understand what you mean by that.
If there's one thing I learned from the OKISO and Liar Macaron incidents, it's that if people catch you making a song with AI your career is essentially over. The former was blatant but the latter was more sneaky and used real UTAU vocals. The reason Liar Macaron remains one of the most watched MVs is because by the time it was exposed as an AI song it was already huge.
No.22293
>As much as it sucks, this "2nd golden age" of Vocaloid will eventually come to an end and bring us back to where we were 10 years ago.I don't think this is really as much of inevitability as you're suggesting. Vocaloid will undoubtedly cool on some timeline, but I wouldn't forecast a similar "collapse" or decay that was experienced then for a number of reasons, mostly because the audience has a larger global footprint and also because there is actual money to be made in being a producer now.
It's easy to imagine any number of scenarios (algorithm tweaks alone can upend entire industries now, let alone subcultures) but I am curious to see how generative AI impacts vocal synth music overall, both on the music itself and whatever identity Vocaloid has in total. The best destiny for Vocaloid is probably to lean on the characters themselves and you're seeing a lot of that again.
>>22291>Better ways to do live performances probablyEverything on the tech front is being led by Vtubers now so I wonder if Vocaloid is really even a part of this conversation any more. I think the better framing is how concept-focused lives like Decomiku and Project Voltage might get more into the mix. Very early days for that stuff still.
No.22302
>>22301It already kinda has
No.22305
>>22292I just meant if the software companies made like a lot of tech companies and sell out their main consumers and fans to chase a hype thinking it would increase revenue.
>>22293Yeah though vtubers / virtual singers have been around quite a while now and while I get your point there still is a sort of niche here because synthesized vocals are capable of things human voices are not and can have a specific unique sound. I more meant how some producers also like to be performers doing live sets and being able to control vocals live gives more flexability than timing live music to a recorded vocal track or using just sections in a sampler.
No.22306
>>22305Also PinocchioP literally sings
No.22308
>>22294I don't think it'll be as bleak as you make it sound
No.22310
>>22306A lot of producers have weirdly good singing voices. I do wonder how they would handle singing duet controlling the vocal synthesis live. I mean in cases of it not being voice changer based.
No.22313
>>22310I have only heard Pinocchio, Crusher, Hachi and Yuyoyuppe sing
No.22327
>>22302Yes, this second golden age was proof of it.
It's why I posted that picture. IA might not be a Superhero anymore, but Teto definitely is, though in the future someone else might be, depending on trends. Vocaloid will continue to exist but it will be in a form we might not recognize.
No.22330
>>22328Aieeee pls delete that can't be IA that has to be an impostor
No.22334
>>22330It's just Luka in IA cosplay
No.22335
>The fact that this site is so active in an era where imageboards are barely relevant anymore goes to show how many people care about Vocaloid.
Vocaloid forums in general are active. Check Vocaverse for example and you'll see that all of the boards are active and with recent posts. Weirdly though, it does seem like 39chan is the most active one right now considering it has a general PPD of 100+ posts (203 today). I wonder if it's just the novelty or the fact that you don't need an account to post here.
No.22350
>>22335Because you don't need to make an account to post or see images
No.22365
>>22294Anon you're overthinking it a bit
No.22383
>>22301Yep, I think it will transcend generations the way some people from my generation appreciate older music from the 60s and 70s
No.22400
>>22335What other people said and also /all/ makes it easy to see the active threads. On some of the vocaloid BBcode forums they have too many subforums for the activity, so you have to go hunting for new posts, and that makes the site feel less lively
No.22413
>>22400More forums should have an overboard like most imageboards do
No.22438
>>22437I don't think AI music will ever truly replace human music
No.22461
>>22400Forums in general often have overboards like imageboards do they just label it different and it isnt as easy to spot. They often make it so you can see by recent post or recent thread.
No.22464
>>22461Vocaverse has one but i kinda prefer how it works on imageboards since i don't have to enter the threads to read them and i can just reply to whatever i find interesting
No.22468
>>22464I agree, but I do wish the number of replies shown was more dynamic. It's always five (except for stickies), regardless of the time between them, so for threads with active discussion, you often have to expand the whole thread just to read the recent posts, while slow threads can have posts separated by months all on the front page.
No.22469
>>22468We can change that.
But it's important to hear what the others think. How many posts would you suggest?.
No.22477
>>22469Not that anon but the thought about changing it dynamically based on thread speed was interesting. Maybe something like trim any posts from over 1 month ago, or add 2 more posts if they’re newer than 1 hour, or something. As a baseline the current number is comfortable though.
No.22479
>>22464I think I have seen some forums do it this way but they dont exist anymore so i dont know what the code or software they used to do it.
No.22481
>>22469>How many posts would you suggest?Not a fixed amount.
Basically what I'm suggesting is a "smarter" script that looks at the recency of posts as well as their proximity in time to one another, in order to decide what shows up on the board. So an old thread with one new reply, would only show that reply; while a thread with an active discussion, would show all the posts starting from the one that began the current conversation, or just the most recent ones if it exceeds some limit.
The idea is that this would be more intuitive for most people, who often don't look at the post date when browsing a board, and so get temporally confused, and may even reply to an ancient post as a result. But this is easier said than done, and should probably be an option that you can turn off if it doesn't work properly.
No.22483
>>22481>if it exceeds some limit.Let's say 10, that might be too much, but I should give a number.
It occurred to me that I must not be the first person to have had this idea, so you might want to find out if someone else has already tried it and see how it went.
Also this really only makes sense for discussion boards, boards like /c/ should not have this on by default. Hiding posts just because they are older than posts surrounding them might be a bad idea in general, and may make the board look devoid of content, so you should carefully consider whether you want something like this to be the default experience, if you go to the trouble of implementing it.
Five posts per thread works well enough, and it may not be possible to improve upon it without trade-offs and downsides.
No.22587
>>22437Thank you..I always find it ironic vsynth producers who use an AI program are against minimal transformative uses of AI. (To clarify, no, I do not support 100% generated kuso like Fruit Love Island or CryDie)
No.22640
>>22630AI isn't necessarily a bad thing. It kinda depends on the use you give it which in most cases tends to be for unethical stuff. Vocaloid, SV, etc all use generative AI but it's to make the voices sound smoother, the process of making a Vocaloid song hasn't changed at all and you still gotta tune the vocals yourself.
No.22642
>>22640Yeah but try explaining that to the average Vocaloid fan
No.22654
>>22640Technically AI is just a marketing term. It really is machine learning.
No.22657
>>22630My question is why don't you need a gazzillion gigs of vram for vocaloid and synthv like you need for gen ai stuff? How did they smooth that out for the consumer?
No.22658
>>22657Training is what takes all of the resources. Using the bank afterwards just calls the algorithm.
No.22660
>>22658 (doubleposting)
For that reason I would still put Vocaloid/SynthV in a different category than most Gen AI you'd think of... it's not like you write a prompt and it spits out the song. You're still programming the output, and the results are basically user-defined. The machine learning just replaces the re-synthesis algorithms of traditional voice synth.
No.22661
>>22658Sure but the few times I tried to toy with already trained stuff, even voices, it usually requires a lot of vram. Even if it’s just me converting some singing. Idk my computer is a hunk of shit so maybe it’s just bullying me.
No.22662
>>22655>forget about the worst AI bubble trashUnless you're banking on the collapse of every UGC and streaming platform, I wouldn't count on it. Even if you approach it narrowly as a backlash to the tech, people already can't distinguish artificial from real anymore with state of the art models. Spotify's charts are being infected with middle of the road generative music today because 1) people broadly do accept it already, whether they care to notice the usage/disclosure or not, and 2) it's cheap to make and infinitely abundant. An AI bubble popping tomorrow wouldn't change those economics, even the cheap part. The power people do have left to resist this is by abandoning the convenience of platforms like Spotify that thrive on churning mediocrity for a profit. Saying this as someone that does have a homelab kitted out with Plex and more: good luck.
No.22673
>>22661Vocal synth editors have a lot more manual input and usually render note-by-note, so I'd imagine that helps. I've heard people complain about SynthV 2 being laggy on low-end PCs, though.
No.22713
>>22671I feel like most iconic commercials use copyrighted music anyway if not memorable original tracks