The train is on and it is picking up its pace. Artificial intelligence (AI) already makes it possible to dub actors in all languages, while retaining their original voices. All of this is done in the name of the artistic integrity of the actors. But to the detriment of a profession that employs nearly 4,000 people in Quebec.
See for yourself. Tom Cruise interviews Jack Nicholson in the French version of A Few Good Men, by Rob Reiner, edited by the artificial intelligence experts at Flawless AI. But it is her voice with its intonations that we clearly recognize… rather well synchronized with the movement of her lips. The result, which can be seen in their promotional video, is remarkable.
This company, which has a storefront in Hollywood and London, was co-founded by English producer and director Scott Mann, who notably directed the films Heist, with Robert De Niro, Final Score and Fall, released in 2022, which is the first film to benefit from TrueSync technology.
Using this software, Scott Mann managed to replace the approximately 30 F*** words of the two lead actresses by re-recording other words over it in post-production. It is through this process – commonly known as deepfake, where the movement of the mouth is altered to accommodate new phrases being spoken – that it earned the ‘PG 13’ rating…
A technology that raises, it must be said, some concerns, and that the American artistic community would like to see framed, since it makes it possible to rewrite the dialogues from scratch, and therefore, in theory, to make the actors say what they never said.
Fall could have been doubled thanks to this same software, but for the moment, TrueSync is still running. Clearly, it is a question of modulating (subtly) the movements of the mouth in order to secure them with the translated dialogues, delivered with the voice of the interpreter – of which the system has archives.
A process that we could appreciate in the documentary The Andy Warhol Diaries (offered on Netflix), which Andy Warhol himself narrates, despite the fact that he has been dead for 36 years! A work carried out by the firm Resemble AI over a period of several months.
Nevertheless. Flawless AI is getting closer to its goal: to end human dubbing in the name of actors’ “artistic integrity”. In an interview with Time magazine, Scott Mann said that the dubbed dialogue is “rarely in sync” with the actors’ facial movements. Second, “they sacrifice meaning and nuance” of dubbed content, he believes.
The American film Every Time I Die, by Robi Michael, released in theaters in 2019, was the first film to benefit from this technology from A to Z. An initiative of the young Israeli company Deephub, which dubbed the film thanks to to the AI in 2022 – using the original voice of each of the actors.
Another project that saw the light of day this year using the same technology was an advertising campaign led by the firm Synthesia to eradicate malaria, which featured footballer David Beckham. Again, the dubbing was done in several languages, with the voice of Beckham.
In Quebec, despite these impressive advances from the viewer’s point of view, there is much more fear than excitement about AI dubbing. Because if – or rather when – the American steamroller really gets going, this profession and this know-how risk disappearing.
According to the Union des artistes (UDA), 44% of the 8,450 active members declare income from dubbing, but also from narration (whether documentaries, audio books or video descriptions), video games or even advertisement. So many sectors targeted by artificial intelligence software. We are therefore talking about more than 3,700 people directly affected.
“Dubbing with artificial intelligence software represents a serious challenge”, confirms the management of the UDA, who intends to meet the Minister of Culture and Communications, Mathieu Lacombe, “in the very short term” . “It is an issue, precisely because it is a sector that employs nearly half of our members, but also because it affects our Quebec cultural identity. With synthetic voices, we lose the specificity of our language. »
Frédérik Zacharek is an actor who does a lot of dubbing. To the point of now being a set director for dubbing projects. He worries about the prospect of his workplace – there are around 750 actors who specifically dub films and series, according to the UDA – being replaced by synthetic voices.
For the past week, a petition has even been posted on Facebook by Frenchman Olivier Barbery, editor-in-chief of the magazine Synchro, specializing in dubbing. As of Thursday, the Doubling petition: against AI to replace actors had collected nearly 7,500 signatures, thanks in particular to the participation of several hundred Quebec signatories.
“Lowering the production costs of dubbing in France and in French-speaking countries such as Belgium or Quebec can encourage the majors and streaming platforms to use AI in this sector, and thus put a whole section of the profession out of work writes Olivier Barbery.
The dubbing specialist urges signatories to “act now”, before it’s too late.
From the viewer’s point of view, not everyone wonders who is behind the speaking voice, but Frédérik Zacharek believes that Quebecers identify with local voices. When actress Béatrice Picard, who was the voice of Marge Simpson for 33 years (in the French version of The Simpsons series), bowed out last January, she received several tributes.
As soon as a film or series needs to be dubbed, the five main Quebec studios get busy – Difuze, Cinelume, Pixcom, La Belle Équipe and Mel’s. The texts are written by adapters, then the actors enter the studio to record their voices.
For example, Alain Zouvi is the Quebec voice of Brad Pitt; Bernard Fortin, that of Tom Hanks; David Laurin, that of Leonardo DiCaprio; Aline Pinsonneault, that of Reese Witherspoon and Natalie Portman; Guy Nadon, Dustin Hoffman; Isabelle Leyrolles, that of Jennifer Aniston and Eva Mendes; Gilbert Lachance, that of Tom Cruise and Johnny Depp.
When the projects of Flawless AI, Deephub, Resemble AI, Synthesia and others multiply, unless a legislative framework is put in place, these voices will die out.
