AI text-to-speech is hailed as the future of video games, but at what cost? | Artificial Intelligence (AI)

AI Video & Visuals


W.When the epic open-world PlayStation 4 game Red Dead Redemption 2 was developed in 2013, it took 2,200 days to read 500,000 lines of dialogue from 700 voice actors for the 1,200 voices in the game. rice field.

This was a feat that would have been nearly impossible for any other studio, let alone a game studio smaller than Rockstar Games.

But advances in artificial intelligence are making it easier than ever to recreate the human voice to create automated real-time responses, nearly limitless interaction options, and voices tailored to user-specific input. However, this technology raises questions about the ethics of speech synthesis.

Australian software developer Replica Studios launched a text-to-speech platform for game developers in 2019. This is the tool used by Australian game developer PlaySide Studios in pre-production for their game Age of Darkness: Final Stand.

“We hope there are thousands, if not hundreds, of other studios dreaming of developing games like this. [Red Dead Redemption 2] Because everyone wants it,” says Shreyas Nivas, CEO of Replica Studios.

Recording every single line of dialogue “is inefficient from a cost standpoint, but also very inefficient from a time standpoint. It also requires such a large team,” says Nivas. says Mr.

The company says Replica has licensed the voices of 120 actors for use in video games, and can represent up to 1,000 different voices.

Nivas sees AI-powered speech synthesis as the future, but like many AI advancements, the practice presents ethical dilemmas.

There are now free online text-to-speech tools that can be used to imitate the voices of celebrities, film and television characters, often without the artist’s permission. And Bloomberg reported earlier this month that some voice actors were “shocked” to find their voices used in content they weren’t part of (they later said the terms of their contracts were such). was wide enough to cover a wide range of uses).

Replica technology demonstration

Earlier this month, in Japan, the Japan Performing Arts Workers Association held a press conference to express its concerns about the impact of AI on voice actors and music. The group, which reportedly has about 52,000 workers as members, called for legislation to protect their jobs.

Nivas said Replica pursued a licensing model that allows actors to earn money from using their voices outside of studio recordings.

“There are many things to improve, but we try to build this work in such a way that we take the voice actors on a journey with us. I am trying hard.”

Stock image of Australian software developer Replica Studios' text-to-speech platform for game developers
Stock image of Replica Studios’ text-to-speech platform. The software developer believes AI will make voices more accessible to smaller game studios, creating more jobs for early-career actors. Photo: Replica Studio

Nivas said the big video game companies will probably continue to hire high-profile talent like Troy Baker, who voices Joel Miller in The Last of Us, but AI will make small-scale jobs easier. Big game studios say voices will be easier to access and jobs for young actors will increase.

Skip past newsletter promotions

The company has also introduced content moderation policies against the use of voice in fraudulent or illegal content. Nivas said the learning process that accompanies AI acceleration is still in progress, but he believes the acceleration should continue.

“There are a lot of unknowns, but the best thing we can do is watch how things go. But keep moving forward.”

The Alliance for Media, Entertainment and the Arts, the Australian trade union representing actors, has yet to develop a policy specifically focused on the impact of AI on acting, but a spokesperson said the promise of AI is for performers. It should not come at the expense of quality and meaningful work for

“Actors must retain ownership of their voice and this should not be permanently transferred by license,” the spokesperson said. “We urge producers to work with performer representatives in this emerging technology space.”

Canadian singer Grimes last month called on people to create new songs using his AI-synthesized voice and said he would split the royalties 50/50.

early this week she tweeted She emphasized that people are “starting to make songs with a grime sound that’s more competitive (or maybe even more??) than mine.”

“But it’s also the most wonderful and poetic way to die and come back in another career.”

In fact, I’m a little stressed that people are starting to make grime-sounding songs that are more competitive (or more??) than I am, but that’s because it dies and comes back in another career. is also the most wonderful and poetic way of

— 𝔊𝔯𝔦𝔪𝔢𝔰 (@Grimezsz) May 8, 2023

This article was amended on May 15, 2023 to clarify that PlaySide Studios used voice synthesizers in pre-production and not in the final game.





Source link

Leave a Reply

Your email address will not be published. Required fields are marked *