Get your ticket to GaijinPot Expo 2024
This week, Koichi Yamadera (the voice of "Cowboy Bebop’s" Spike Spiegel), Romi Park ("Fullmetal Alchemist’s" Edward Elric), Ryusei Nakao ("Dragon Ball’s" Frieza), and 23 other prominent anime voice actors and actresses announced a new project they’re collaborating on. If that sounds like way too large a cast for an initial cast reveal for a new anime project, you’re right, because they weren’t announcing a new animated TV series of theatrical feature, but instead a collective stance against unauthorized use of performers’ voices by generative AI.
The awareness project, which is called “No More Unauthorized Generative AI” (or “No More Mudan Seisei AI,” in Japanese), has posted a video to its YouTube channel, with Nakao making the opening statement.
“Someone was selling my voice without permission. I was shocked. Our voices are our livelihoods. They are our lives. Please listen to how it makes us, as voice actors, feel when our voices are used without permission for generative AI,” says Nakao, followed by the 26 members of the project shouting in unison “No more unauthorized Generative AI!”
Those feelings are expanded upon in a press release from the group, which says:
"Readings and songs that we have no memory of recording, and our voices themselves, have been uploaded to the Internet, sometimes being offered for sale. Our voices are our livelihoods. They are our lives, an important part of who we are that we have grown up with. Even if these uploads are coming from fans who want to hear more of our voices, it brings us no joy for our voices to be used without permission.
"New technologies are likely to bestow great benefits upon humanity in the future, but at the same time, we want all of us, together, to broaden our perspective to consider each other’s feelings, what kind of culture that future will have, and discuss how those technologies will be used. We have created this video as a way to start that process.
Rather than hurtful words and retaliation, we hope to create cultural rules through peaceful discussions with the involvement of experts in order to see eye-to-eye, to protect the fertile soil in which good works can be created for the next 10, the next 20 years."
No More Unauthorized Generative AI’s statement comes less than a month after Aoni Production, one of Japan’s highest-profile talent agencies for anime voice actors and actresses, said it will be entering into a partnership with AI company CoeFont to create AI versions of the voices of a number of its performers, including the voice of "Dragon Ball’s" Goku, Masako Nozawa, for applications in “non-acting” projects. It’s worth noting, though, that No More Unauthorized Generative AI’s message is centered on unauthorized (and by extension non-compensated) use, and isn’t being framed as a blanket indictment of replicating performers’ voices using AI.
The complete list of voice performers currently part of the project consists of Michihiro Ikemizu, Yoji Ueda, Yuko Kaida, Yuki Kaji, Tomie Kataoka, Mika Kanai, Kujira, Shuhei Sakaguchi, Chika Sakamoto, Shunsuke Sakuys, Yuko Sasaki, Bin Shimada, Yu Shimamura, Tarusuke Shinkagi, Toshihiko Seki, Ryota Takeuchi, Hiroki Touchi, Ryusei Nakao, Joji Nakata, Daisuke Namikawa, Romi Park, Rika Fukami, Juna Fukuyama, Katsuhisa Hoki, Mitsuru Miyamoto, and Koichi Yamadera. The members plan to release individual videos expressing their personal feelings on the unauthorized use of voices for generative AI as well.
Source: YouTube/ NOMORE無断生成AI, Otaku Kenkyu via Yahoo! Japan News
Read more stories from SoraNews24.
-- Goku’s voice actress’ voice to be converted to AI for “non-acting” projects in Japanese, English
-- Japan has more anime/video game actresses than ever before, according to major seiyu magazine
-- Anime voice actress launching new web series on the wonderful world of sake 【Video】
© SoraNews24
10 Comments
Login to comment
sakurasuki
Some people voice can really similar to many people. So it doesn't AI necessarily stole from them.
https://www.scientificamerican.com/article/human-voices-are-unique-but-were-not-that-good-at-recognizing-them/.
Jonathan Prin
A lost war from the start.
factchecker
Too late.
owzer
I have no idea how they’re going to be successful in combating this. As early posters I’ve already pointed out, you can mimic another person’s voice that sounds similar to the voice actor. Where do you draw the line? I think it’s simply too late to do much about it.
People should be more concerned about scams in the future that mimic their voices and trick even the most careful of people into disclosing information and spending the money where they shouldn’t.
YeahRight
Yeah. Good luck with that.
Toblerone
Unfortunately, the writing is on the wall. Entertainers and artists are going to be hugely disadvantaged.
Abe234
Next step, it’ll be going after the computer programmers job! Doing yourself out of a job or dumbing your skills down.mmmmm not sure which. like the computer before, some jobs will disappear and some new ones will re-appear, just gotta accept their will be some lag between the two. Nobody saw YouTube or Tim Tok or Amazon when the internet was created. But someone did! Remember only humans pay tax to pay for the services we use. So they won’t take all our jobs. Some will always and forever need a human. Empathy, trust, distrust, warmth, touch, imagination, will always be the domain of humans. I wonder if AI will be like the old computer programmer, garbage in garbage out. Mmmmmm!
Jonathan Prin
Actors and actresses have to play live and get paid for that.
Life has been too easy to get several successes for one effort.
When I make a report, I am paid once, not depending on the use of the report. For most of us workers, that was the way and should be the way.
JRO
It's not mimicking or just sounding like their voices, these models are fed the data containing their voices without any rights to that material, gAI has it's uses but the rights holders of the data they want to train on needs to be paid, crazy to think otherwise, it's like wanting to bake without paying for the ingredients. Without any restrictions and changes to copyright law no one can live or work as a creative. As these things get better I will just be able to copy paste a movie or song and tell it to make it different enough to pass as original content. All commercial gAI should need to have their training data public and get the rights for any content they use.
Kazuaki Shimazaki
I'm going to go my usual route of supporting the AI. It might suck for the content creator, but AI is a national security imperative and limiting yourself from exploring its full possibilities (unless we can somehow get every significant actor to restrain themselves) is to shackle yourself in the tech race.
I really don't get that argument. Fundamentally, it's doubtful if ANY human creator got to where they were without acquiring samples from their predecessors. In the case of voice actors, they have been watching anime from when they were kids voiced by other voice actors, and even if you are the First Generation you must be exposed to theatre voice work.
I'd be surprised if any one of them managed to completely avoid referencing any of those memories when doing their own voice work. (And of course, they didn't buy the rights either)
[If anything, the AI is no different from a slow, inefficient learner - in that it requires many times the sampling of a human to crank out a usable product]