“The day has come: I have been replaced by a voice generated by artificial intelligence,” tweeted Alejandro Graue, a 36-year-old Argentine voiceover and dubbing actor. For the past few months, he had been dubbing into Spanish the content of a famous YouTube channel with millions of subscribers that he prefers not to name. “The speaker is histrionic and uses a lot of particular idioms. He sometimes stutters, stops in the middle of a sentence, and starts talking about something else. To dub this well, you need some acting technique,” he told EL PAÍS.
In January, Graue saw the channel was being updated with videos that he had not dubbed. “When I hit play, I heard a monotonous voice saying, ‘Hello, welcome back to this program, today we will be...’, or something awful like that,” he said. People complained about it in the comments, but new videos with the robotic voice dubs continue to appear.
Graue’s tweet has been viewed by over 740,000 people. He’s not so concerned about a lost gig – he has others – for now. The problem lies ahead. “It’s concerning. The human factor may soon disappear from creative expression,” he said.
In just a few months, the voiceover industry has been upended by rapid advances in artificial intelligence (AI) technology. The explosion of ChatGPT and applications that generate photos and illustrations based on text prompts has reached the video and film voiceover world. AI machines still can’t do everything in the creative vocations, but there is no human light at the end of this road. “Our workload has been steadily dropping for the last two or three months,” said Noemí Gutiérrez, director of the Voces en la Red dubbing company. “All our actors have been contacting us because they’ve noticed the drop as well. There are many more platforms now that provide free synthetic voices. Some people are happy with a crappy voice and they just go with that,” she said.
Begonya Ferrer, a dubbing actress, says some haven’t yet felt the full impact but can sense what’s ahead. “Since I work for many people in various countries, I don’t know if I’ve been replaced by AI somewhere,” she said. But a certain type of work has been cut back. “They use artificial voices to do a first pass or voiceover draft, and then have the voiceover artist do the final dub,” said Ferrer. “This way, perhaps, they avoid paying for two voiceovers.”
Quality and nuanced intonations are still important in this field, but many are sure that the AI capabilities will continue to improve. “I understand that it’s technological evolution and that artificial intelligence can serve many needs like answering machines and recorded messages on subways. There’s no need for feeling in those voices. But I don’t think it should be used for anything that involves acting, even if it’s just to preserve the profession,” said Graue.
Only programmers needed
In more and more professions, humanity will be in the hands of people who can talk to machines, says Graue. “Having a human being execute a task will not be necessary. Only programmers will be needed, and that’s it,” he said. In the comments to Graue’s tweet, some people told him to forget about “preserving the profession,” that AI is like electricity. “People have compared it to the discovery of electricity and how the gas lamplighter job disappeared. I disagree because electricity has global benefits for people. What’s in it for the public, this synthetic voices thing? It only benefits the company that buys it,” he said.
Alejandro Graue isn’t old enough to retire soon and forget about it all. “Maybe if I were 70, I wouldn’t be so worried about the future. But this will affect me directly. I can go into another field, but I must always think about doing work where I can’t be replaced. I’ve talked to colleagues, journalists and screenwriters – everything is moving to artificial intelligence.”
That rapid spread of AI to many professions and vocations leads to anxiety. It’s hard to be calm when machines do tasks in seconds that once required special skills and years of study or apprenticeship.
“I had been watching all this on social media, especially with illustration and graphic design. Then it happened to me,” said Graue. “That’s when I began paying closer attention. Of course, there is fear. Some say it will only replace people who do poor work, and those who do high-quality work will always have jobs. But others are already looking for something else. There’s a bit of everything. Personally, I believe it [AI] will gain ground, little by little.”
Graue is confident that help will come from the law, not a company. “I hope there will be some kind of regulation. I understand that in Argentina, at least, some regulatory work has begun. I hope some sectors won’t be allowed to replace the voices of speakers and actors with machines.”
Brad Pitt with a regional accent
Some companies in the dubbing and voiceover industry are searching for ways to stem the loss of business with technologies that preserve a voice actor’s work. Tools exist that can easily clone or create a timbre, which is a specific vocal sound. But it’s difficult to specify rapidly changing pitches and intonations to express surprise, emotion and anger in two short sentences. “You can generate a voiceover of Brad Pitt speaking with a Welsh or Andalusian accent, as long as you feed the system a prosody [the rhythm, intonation, stress and related attributes of speech] with those accents, created by an actor,” said Javier de Alfonso, founder of Voces en la Red.
“Everything is speeding up – six months now seems like an eternity to me,” said de Alfonso, who is taking a course to adapt to these changes in his business. “We are learning and attempting to serve agencies with very compressed recording schedules. We have tested several platforms, including a very advanced one that has more buttons than a 747.” It’s hard to escape the feeling that these moves are like rearranging the deck chairs on the Titanic.
This tenuous technological lifeline can already enable dubbing an entire film with just two people, while maintaining the actors’ original timbre. “It’s already disruptive,” said Noemí Gutiérrez, director of Voces en la Red. “The system changes the voice of the dubber and gives it the same timbre of the original actors. So, one man can dub all the male voices, and one woman can do all the females and the children’s voices as well.”
“Synthetic voices still cannot get all the emotions right,” said Gutiérrez. “That will take time.” When the technology is fully developed, one can envision voice actors selling their personal timbres, while others with less sonorous voices will sell their acting ability for artificial intelligence to replicate.
Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition