I'm going to be taking just a couple commissions! DM if interested!

1458. It’s embarrassing to show an unfinished drawing.

I rather take health advice from my grandfather who has had two heart attacks and is still healthier heart wise than this guy

「装甲騎兵ボトムズ」

Cured. Seven people. Fucking *cured*. Incredible.

Trust your feelings, Got to live and learn!

A rectangle is walking :3

ガンダム

Allentown native Ian Riccaboni has been named Fightful Announcer of the Year after a standout year on ROH and AEW commentary. A national honor for one of wrestling’s most respected voices! https://lehighvalleywithlovemedia.com/lehigh-valleys-ian-riccaboni-wins-fightful-announcer-of-the-year/

mallow put on some youtube poops on the hotel room's TV during the final night, and we all started howling with laughter over this whole segment

Unused Eggman animation from Knuckles' Chaotix. It's not known for sure when this would have appeared, possibly in level six or seven.

"The Advantage to being a Mexican with no searchable English online history." OH ABSOLUTELY. This applies to AEW too btw. MBT being an edgelord is nothing comparing to known femicide apologist Rush. Same with Dralistico. But it didn't happen in gringolandia so I guess we just ignore it.

A thread on AAA... 🤼

This is canon Metroid

you can say that Miku will, in fact, be AI soon, because she has an AI voicebank due to release this year -- but that doesn't really change anything about the way Hatsune Miku has existed since 2007, does it? still a singing robot that works pretty much the same way

- listen to a machine, and that's why Vocaloid just became its own niche, instead. it's a culture interested in the craft of vocal synthesis specifically, which engages with music differently from the average person. stuff like SynthV is just another chapter in the history of this particular culture

the final concern could be "ok, but is it replacing real singers?" and like, you can argue that you CAN, in theory, use SynthV instead of a human singer, but that same concern has existed for the last 20 years since VOCALOID was created and it... hasn't really happened? most people don't want to -

3. it's not making art /for/ you. you can't just ask SynthV to generate you a song (unlike other AI models like SUNO), you still have to input notes and lyrics by yourself. this workflow is the exact same workflow that has existed for the past 20 years

2. they don't need data centers worth of computing power and gallons of water to work. in the video above, SynthV was running entirely within my computer, and it's in fact less demanding than the average videogame

there have, in fact, even been situations in which a singer (KAF) decided to revoke consent to the voicebank being made, and the company had to comply and cancel production. bare minimum, but miles above the practices of OpenAI, for sure

1. AI voicebanks aren't made with huge batches of data scraped from the internet, but with the voice samples of voice providers who are hired, paid, consent to their voice being used for training, and sometimes get royalties off of VB purchases (Una above, for example, is the voice of Aimi Tanaka)

it's scary calling vocal synthesizers Generative AI because Generative AI is what's been culturally decided is the evil kind of AI, because see, most current vocal synths Are Generative AI but they also just don't share the same ethical concerns as stuff like Chat GPT:

concat voicebanks are usually multiple GBs worth of data, because they have to include all the original sample files. AI voicebanks tend to be under 500 mbs /because/ they don't have the original sample files, just the model

something you may see around is "the AI is more similar to a feature that adds pitch and timing variation to make the singing sound more realistic", and while it /does/ do that, it isn't doing that on top of sliced samples. the audio is, in fact, entirely generated

for the end user the process is the same, you input notes and lyrics, but what the program is doing behind the scenes is different. instead of stitching together existing samples, it generates a sound based on what it "learned" and the phoneme/pitch it's been given (here it is with Una in SynthV)

instead of recording specific syllables and phonemes, the voice provider is recording singing normally. those singing samples are then manually labeled and both the singing and labeling data are fed into a training model, which "learns" how it tends to sounds like when the VP sings each thing

if you've ever heard a Miku song, it was probably made using one of her V2, V3 or V4x voicebanks, which are all concatenative synthesis, which was the dominant vocal synthesis technique up until like 2021. not nowadays. current engines like SynthV and VoiSona use AI synthesis, instead

a person is recorded singing syllables in specific pitches, and then in the program when the user inputs notes and lyrics, the engine pitch corrects and mashes those samples together to match the notes that the user input. here's concat in action in UTAU, with Rouon Aro's samples vs splicing

vocal synthesizers have existed for a while, and the VOCALOID series of software (which is not the only one) has existed for around 20 years now. the way they work is with what is called "concatenative synthesis", which is basically splicing together samples of real human singing

good thread articulating the whole relationship vocaloid has with genAI

HE IS VERY DIE HARD! HE’S JUST JAPANESE ROCK ROCK ’N’ ROLLER! HIS REAL NAME IS MASARU TOBITA. KNOCK OUT ALL ENEMY! ATTACK ALL MONSTERS!

watching another post spreading misinformation about the "is Miku AI" debacle getting numbers but instead of fighting in the comments I'm just gonna make my own post so. is vocaloid gen AI?? short answer: it depends, sometimes, and yes in the way you're thinking, but also not really long answer:

[NEW VIDEO] Can AI images be used for comics? I look at an award winning AI manga, how real authors are using AI, and even try an AI tool myself to rediscover why I draw. #art #videoessay www.youtube.com/watch?v=7PrL...

also you still have to, you know write the goddamn song you can't just tell vocal synth programs "make me a song about X" and have it shit something out, you still have to input the lyrics and melody, you still have to manually tune the voice, it's all a human touch at the end of the day

lrp yes i know recent vocaloid stuff and synthV does use genAI technology but from what i understand it's all done using models trained only by the software they already have the rights to and properly compensated the voice providers for. and the resulting carbon footprint is negligible.

Apparently the AI bros are on another round of "Hatsune Miku is AI, checkmate luddites" so this is just a reminder that the singer behind Miku was paid and still earns royalties, and beyond that, Miku is nothing more than a MIDI soundboard with advanced pitch converter options, it is not genAI.

A reminder that US lawmakers and paypros want to effectively criminalize consensual sex work or the ability for artists to draw NSFW art but I'm sure they're completely fine with stuff like this being advertised to kids on Youtube

WWE AAA put up an AI generated video for Dom Mysterio and the slop machine made him AEW Champion lmao

Taking all my fitness advice from somebody who was on the juice so hard that he needs a machine to keep his heart operational in his 50s.

went out and touched grass today!!

No more chicken!? What!?

i got it thank you!!!!

Uni and Mini Uni

Three years ago today we sadly lost Jamin Pugh. Known worldwide as Jay Briscoe, he was a star in ROH for over 20 years, capturing the ROH World Championship twice, and along with his brother Mark, dominated the tag team division winning the ROH World Tag Team Championship a record setting 13 [1/2]

Let's think positive! 🩷 #mayaamano #persona2 #persona #ペルソナ2 #天野舞耶

🫶💜 #inART

SONIC THE HEDGEHOG CD #SONICTHEHEDGEHOG #SONIC