Bourdain Documentary’s Use of A.I. to Mimic Voice Attracts Questions

Bourdain Documentary’s Use of A.I. to Mimic Voice Draws Questions

The new documentary about Anthony Bourdain’s life, “Roadrunner”, is an hour and 58 minutes long – much of it is filled with footage of the star from decades of his career as a celebrity chef, journalist and television personality.

But on the movie’s opening weekend, 45 seconds of it gets a lot of attention.

The focus is on a few sentences of what an ignorant viewer would mistake for prerecorded audio of Bourdain, who died of suicide in 2018. In reality, the voice is generated by artificial intelligence: Bourdain’s own words, converted to speech by a software company that had received several hours of audio that could teach a machine how to mimic its tone, cadence and intonation.

One of the machine-generated quotes is from an email Bourdain wrote to a friend, David Choe.

“You are successful and I am successful,” says Bourdain’s voice, “and I ask myself: are you happy?”

The film’s director, Morgan Neville, explained the technique in an interview with Helen Rosner of the New Yorker, who asked how the filmmakers could possibly have gotten a shot of Bourdain reading an email to a friend. Neville said the technology was so compelling that viewers probably wouldn’t realize which of the other quotes are fake, adding, “We can have a documentary ethics panel about it later.”

Now seems to be the time for such a panel. Social media has erupted with opinions on the matter – some find it creepy and tasteless, while others are unmolested.

And documentary film experts, who often grapple with ethical issues in feature films, are sharply divided. Some filmmakers and academics see the use of sound without disclosure to the audience as a breach of trust and a slide in the use of so-called deepfake videos, which contain digitally manipulated material that appears authentic.

“It wasn’t necessary,” says Thelma Vickroy, chair of the Department of Cinema and Television Arts at Columbia College Chicago. “How does the audience benefit? You conclude that he said that during his lifetime. “

Others don’t see it as a problem, considering that the tone comes from Bourdain’s words and an inevitable use of evolving technology to give voice to someone who is no longer there.

“Of all the ethical concerns you can have about making a documentary, this seems trivial,” said Gordon Quinn, a longtime documentary filmmaker known for executive producer titles such as Hoop Dreams and Minding the Gap . “It’s 2021 and these technologies are out there.”

Using archive footage and interviews with Bourdain’s closest friends and colleagues, Neville examines how Bourdain became a worldwide personality and examines his devastating death at the age of 61. The film “Roadrunner: A Film About Anthony Bourdain” received positive reviews: A film critic for the New York Times wrote: “With immense perception, Neville shows us both the empath and the narcissist” in Bourdain.

In a statement on the use of AI, Neville said Friday that the filmmaking team had received permission from Bourdain’s estate and literary agent.

“There were a couple of sentences Tony wrote that he never spoke aloud,” Neville said in the statement. “It was a modern storytelling technique that I used in a few places where I felt it was important to bring Tony’s words to life.”

Ottavia Busia, the chef’s second wife, with whom he shared a daughter, appeared to criticize the decision in a Twitter post, writing that she hadn’t given filmmakers permission to use the AI ​​version of his voice.

A spokeswoman for the film did not immediately respond to a request for comment on who gave permission to the filmmakers.

Experts point to historical reenactments and voice actors reading documents as examples of documentary techniques that are widely used to provide viewers with a more emotional experience.

For example, documentary filmmaker Ken Burns hires actors to set long deceased historical figures to music. And the 1988 documentary “The Thin Blue Line” by Errol Morris sparked controversy among film critics as it re-enacted the events surrounding the murder of a Texan police officer; the film received numerous awards but was banned from Oscar nominations.

But in these cases it was clear to the audience that what they saw and heard was not authentic. Some experts said they thought Neville would be ethically clear if he somehow disclosed the use of artificial intelligence in the film.

“When viewers start to doubt what they hear, they will question everything about the movie they are watching,” said Mark Jonathan Harris, an Oscar-winning documentary filmmaker.

Quinn compared the technique to one used by director Steve James in a 2014 documentary about Chicago film critic Roger Ebert, who when the film was made could not speak after losing part of his jaw in cancer surgery . In some cases, the filmmakers used an actor to convey Ebert’s own words from his memoir, or they relied on a computer to speak for him as he keyed in his thoughts. But unlike “Roadrunner” it was clear in the context of the film that it wasn’t Ebert’s real voice.

For some, part of the discomfort about the use of artificial intelligence is the fear that deepfake video may become increasingly ubiquitous. At the moment, viewers tend to automatically believe in the veracity of audio and video, but if the audience has good reasons to question this, it could make people plausible to disavow authentic footage, said Hilke Schellmann, filmmaker and Assistant Professor of Journalism at New York University writing a book on AI

Three years after Bourdain’s death, the film aims to help viewers understand both his virtues and vulnerabilities and, as Neville puts it, “reconcile these two sides of Tony.”

For Andrea Swift, film director at the New York Film Academy, the use of AI in those few film clips overtook a deeper appreciation of the film and the life of Bourdain.

“I wish it hadn’t happened,” she said, “because then we could focus on Bourdain.”

Christina Morales contributed to the coverage.