‘Roadrunner’ Director Deepfaked Anthony Bourdain’s Voice

Director Morgan Neville recently revealed in an interview with The New Yorker that the filmmakers used deepfake technology in the production of Roadrunner: A Film About Anthony Bourdain to create a voiceover that sounds like it was done by the deceased chef, author, and documentarian.

In Roadrunner, the viewer watches as Bourdain struggles with his meteoric rise from a normal restaurant chef to a world renowned author and television personality. In one scene, he appears to read part of an email that he sent to a friend aloud. “You are successful, and I am successful, and I’m wondering: Are you happy?” his voice says in the film. 

The New Yorker reporter Helen Rosner asked Neville how the filmmakers got that recording of Bourdain reading the email aloud. Neville explains that this was just one of three written quotes that he wanted Bourdain's voice for but had no recordings of. They sent hours of recordings of Bourdain's voice to four software companies to create an AI model, or deepfake, of Bourdain’s voice before settling on one. Neville did not disclose which companies they used to create the deepfake, or what other portions of the movie used this AI-generated voice.

Motherboard previously did a podcast episode of CYBER in which we went through the process of deepfaking our podcast host (with consent) to show how far the technology has come along in the last few years and the potential issues associated with this tech.

The recent popularity of technology that can accurately simulate someone's likeness or voice has raised numerous ethical quandaries since their inception as a way to own women's bodies by creating porn, which Neville seems to be aware of. “We can have a documentary-ethics panel about it later," he told The New Yorker. Deepfake audio has also been used by hackers to socially engineer employees at corporations.

The Academy Award-winning director told GQ magazine that he asked Bourdain's widow Ottavia Busia and literary executor. “They were like, Tony would have been cool with that. I wasn’t putting words into his mouth. I was just trying to make them come alive.” He even reveals that he contemplated narrating the whole film with his voice using this deep fake technology, but "I stopped myself instantly." 

The reaction to the news online has been largely negative, with some arguing that deepfakes go beyond line reads from other actors. Film critic Sean Burns tweeted that he was not aware of the use of the technology when he wrote his review, adding, "I feel like this tells you all you need to know about the ethics of the people behind this project."

Deepfakes have reached new levels of realism in recent years as people have used them to impersonate government officials and famous celebrities. Using machine learning like this in a documentary, even if it is for a deceased subject to orate their own words, sets a new precedent for films and will undoubtedly add fuel to the fiery debate over the technology.