Survival action flick “Fall” dodged an audience-limiting R rating by using voice-over tech developer AI Flawless to synthesize the sound and mouth movement of softer curses. This director and “Fall” co-writer Scott Mann is ideally co-CEO of Flawless, but replacing more than 30 harsh swear words and scoring the coveted PG-13 rating without costly reshoots is likely to appeal to indie filmmakers with less direct ties. with technology.
“Fall” tells the story of two young women facing a 2,000 foot fall from the top of a lone radio tower. Swearing while acting in this environment probably came naturally. Lionsgate signed on to distribute the film in the United States, but wanted to refine the coarser language to comply with MPAA rules for PG-13 ratings. Mann called on Flawless’ AI TrueSync to fix the problem. TrueSync’s AI created voice and mouth models of the film’s stars, Grace Caroline Currey and Virginia Gardner, generating the sound and appearance of them saying “creepy” instead of dropping F-bombs. re-dubbed audio and reworked visuals are smooth enough that the cast claims they can’t tell when it’s happening, according to a behind-the-scenes interview. The whole process took two weeks and cost far less than the millions of dollars over several weeks that reshoots would have required.
Synthetic media in mainstream entertainment is on the rise as creators become more comfortable and familiar with the technology, though few have such a short distance to travel as Scott Mann. A random sampling of recent examples can cover a wide range of entertainment outlets. Andy Warhol’s Diaries The documentary was narrated in the digitally cloned voice of Andy Warhol courtesy of Resemble AI Aloe Blacc worked with Respeecher and Metaphysic to combine synthetic speech and video, respectively, to produce his multilingual music video “Wake Me Up” . Respeecher also synthesized the voices of Darth Vader and a young Luke Skywalker for the latest star wars TV shows, while Metaphysic debuted a synthetic Simon Cowell at America’s Got Talent.
Deepfake Queen Elizabeth II will deliver an alternate speech and dance on Channel 4
Watch a Virtual Performance of Simon Cowell for “America’s Got Talent”
Samsung’s new MegaPortrait generates Deepfake videos from still images