Voice actors tackle AI: ‘Don’t steal our voices’

The rise of generative artificial intelligence, capable of creating synthetic voices barely dissimilar to human voices, threatens to displace voice artists, voice actors and audiobook narrators who, ironically, every day fuel this technology that could take away their livelihoods.

“We’re fighting a very big monster,” says voice actor and host Mario Filio, whose creativity was immortalized on the animated film’s soundtrack. Madagascar with its catchy chorus “I want to rock the boat! I want to move the boat!”.

The song’s original verse and title was “I like to move it”. But Filio, who dubbed the voice of the partying lemur King Julien in Latin Spanish, and the film’s musical manager came up with the adaptation, which hit.

This Mexican, who has voiced Will Smith and characters like Obi-Wan Kenobi (Star Wars), Winnie the Pooh or Miss Piggy, claims he never received any royalties for that success. But that’s a minor problem compared to the challenge of generative AI, which creates text, images, videos or voices using existing content without human intervention.

To fight this battle, under the slogan “Don’t steal our voices”, some twenty trade unions and trade unions from Europe, the United States and Latin America have created the Organization of United Voices (OVU), which promotes legislation to promote AI and human harmonize creation.

The “arbitrary and unregulated” use of AI could extinguish an “artistic legacy of creativity (…) that machines cannot generate”, warns OVU.

human rights

Voice actors already competed with Text To Speech (TTS), a system that announces texts, with robotic diction, and is used in assistants such as Alexa or Siri.

But AI added “machine learning”. (machine learning), with which one software you can compare one voice sample against millions of existing ones, identify patterns that generate a clone.

“It’s fed with voices that we’ve been uploading for years,” explained Dessiree Hernández, president of the Mexican Association of Commercial Announcers.

“We’re talking about the human right to use the voice and interpretation without your permission,” he adds.

Platforms like revoicer.com offer a wide range for $27 monthly payments, a fraction of what professionals would charge. On its website, it clarifies that “it is not intended to replace human voices,” but rather to offer a cost-effective alternative.

While tech companies continue to hire interpreters, they suspect it’s just to feed their files, and are looking for tools to track their voices in the face of sophisticated piracy.

They are calling for laws that prevent their voice recordings from being used to train AI without their approval and are imposing “human work quotas,” said Colombian broadcaster Daniel Söler de la Prada, who has been lobbying OVU at the United Nations and the World Organization for Human Rights. intellectual property led. .

Mexico, a mecca for dubbing in Latin America, also introduced a bill to regulate this technology. Meanwhile, in Argentina there is already a law that limits locution to persons with a title. And a machine is not, notes Fernando Costa, who fights against the slogan “Don’t use more announcers, don’t spend more”, of the Argentine Union of Announcers and Communicators.

Unlimited

But AI opens up endless possibilities. For example, in the future, Will Smith’s real voice could be heard in several languages, but with the intonation of a dubbing artist, says Filio, after speaking with industry executives.

It doesn’t sound bad when there are jobs and the public wins, “but we have to pay what’s fair,” says Filio, denouncing the “lack of protection” of a union that works independently.

AFP contacted six synthetic speech service companies, but they did not respond to a request for comment. However, he observed a contractual clause stating that the transfer of rights “includes means and methods which do not exist or are not known (…) and may occur in the future”, which interpreters consider to be “abusive” are considered.

Maclovia González, a Mexican broadcaster for well-known brands, is negotiating with an AI company whose name is not known. He has asked many questions so as not to risk the bills on which he lives if he signs, but has not obtained complete information other than a promise of royalties.

Since she was contacted five months ago, other announcers have been hired. “I want to be part of this revolution, but not at any cost.”

“He Has No Soul”

At Art Dubbing, where they dub Christian content, the alarm bells also went off after four customers requested quotes to use synthetic voices.

The founder, the Mexican Anuar López de la Peña, now faces a dilemma: “I adapt or I disappear”, although he does not want to sacrifice human talent.

Filio stopped recording for many clients because he refused to give up “everything”. “It’s time to support my colleagues,” he says, sure that AI won’t be able to displace humans because it simply “has no soul”.