Roadrunner Documentary Uses Deepfake Audio of Anthony Bourdain

The Use of AI to Recreate Anthony Bourdain's Voice in New Documentary

The documentary Roadrunner: A Film About Anthony Bourdain is scheduled for release in US cinemas on July 16th. Like many films of this genre, it utilizes existing archival materials – interviews and unaired footage – to construct a narrative centered around its subject.

Notably, the film incorporates vocalizations from Bourdain that were never previously recorded. Despite this, audiences will hear his voice delivering these lines.

AI Voice Replication Process

Director Morgan Neville revealed to The New Yorker that he sought to include three specific quotes narrated by Bourdain, for which no recordings existed. To achieve this, he employed artificial intelligence to recreate Bourdain’s voice.

The process wasn't straightforward. Neville detailed in a separate interview with GQ that he consulted with four different companies before selecting one capable of the task. This chosen company then processed approximately twelve hours of audio data to train an AI model.

A significant challenge involved pinpointing the precise vocal tone desired. Bourdain’s narration style evolved considerably throughout his television career, requiring careful calibration of the AI.

Ethical Considerations and Disclosure

While less deceptive than some applications of AI and deepfakes, the ethical implications of this technique remain a point of discussion. Currently, the film does not explicitly disclose the use of AI in replicating Bourdain’s voice.

Neville acknowledged this in The New Yorker interview, stating that viewers likely wouldn't identify the AI-generated lines. He suggested a potential ethics discussion could follow the film’s release.

According to Neville’s conversation with GQ, Bourdain’s family expressed their belief that he would have approved of this approach. Neville’s intention was simply to “make [the quotes] come alive.”

Note: This article was originally published on Engadget.