Why does the cloning of Anthony Bourdain's voice in the documentary "Roadrunner" make many uncomfortable?
The discovery that a documentary director used voice-cloning software to mimic the late chef Anthony Bourdain speaks words he never spoke has sparked outrage and raised ethical questions about the technology's use.
"Roadrunner: A Film About Anthony Bourdain," which premiered in theatres on Friday, is based on genuine footage of the renowned celebrity chef and globe-trotting television personality before his death in 2018. However, Morgan Neville, the film's director, informed The New Yorker that a sample of speech was generated using artificial intelligence.
This has reignited discussion about the future of voice-cloning technology, not just in the entertainment industry but also in politics and a burgeoning business sector dedicated to converting text into natural-sounding human speech.
In a blog post published Friday, Andrew Mason, the creator and CEO of voice generator Descript, stated, "Unapproved voice cloning is a slippery slope." "It won't be long until everything goes once you get into a society where you're making subjective judgment judgments about whether certain instances can be ethical."
Prior to last week, the public debate about such technologies centered on the fabrication of difficult-to-detect deep fakes utilizing simulated audio and/or video, as well as their ability to spread disinformation and foment political strife.
However, Mason, who previously created and ran Groupon, claimed in an interview that Descript has frequently turned down requests to restore a voice, including those from "people who have lost someone and are mourning."
"It's not that we want to pass judgment," he explained. "All we're saying is that you need to draw some clear lines between what's acceptable and what's not."
According to Sam Gregory, program director of Witness, a charity that uses video technology for human rights, angry and uncomfortable reactions to voice cloning in the Bourdain case reflect expectations and concerns of transparency and permission. He believes that obtaining consent and exposing technological wizardry at work would have been more acceptable. Instead, viewers were taken aback — first by the audio forgery, then by the director's apparent disregard of any ethical concerns — and vented their discontent on social media.
"It also touches on our anxieties of mortality and thoughts about how someone may take control of our digital image and force us to speak or do things without our consent," Gregory explained.
Neville hasn't revealed what technology he used to mimic Bourdain's voice, but he did say he utilized it for a few words Bourdain penned but never uttered aloud.
"We employed AI technology with the consent of his estate and literary agency," Neville stated in a written statement. "I utilized a modern storytelling style in a few instances where I believed it was necessary to bring Tony's words to life," says the author.
Neville also told GQ magazine that he received Bourdain's widow's and literary executor's consent. "I surely was NOT the one who stated Tony would have been okay with that," Ottavia Busia, the chef's wife, reacted through Twitter.
Although Microsoft, Google, and Amazon have dominated text-to-speech research, a number of companies, such as Descript, are also offering voice-cloning software. From talking customer service chatbots to video games and podcasts, there are a variety of applications.
On their websites, several of these voice cloning firms prominently display an ethical policy that describes the rules of usage. Many of the companies contacted by The Associated Press claimed they did not replicate Bourdain's voice and would not have done so if requested. Others were silent.
"We have fairly strict restrictions surrounding what can be done on our platform," said Zohaib Ahmed, the founder, and CEO of Resemble AI, a Toronto-based business that provides a bespoke AI voice generator service. "Creating a vocal clone necessitates the agreement of the person whose voice is being cloned."
Ahmed said the few times he's permitted posthumous voice cloning were for academic studies, such as a project using Winston Churchill's voice, who died in 1965.
A more typical commercial application, according to Ahmed, is to edit a TV commercial with genuine voice actors and then personalize it for a certain location by adding a local reference. It's also used to dub anime movies and other media, according to him, by taking a voice in one language and making it speak another.
He likened it to other entertainment industry advancements, such as stunt performers and greenscreen technology.
According to Rupal Patel, a professor at Northeastern University who runs another voice-generating company, VocaliD, that focuses on customer service chatbots, just seconds or minutes of recorded human speech can help teach an AI system to generate its own synthetic speech, though getting it to capture the clarity and rhythm of Anthony Bourdain's voice probably took a lot more training.
"If you wanted it to sound exactly like him, you'd need a lot of data, maybe 90 minutes," she added. "You're training an algorithm to mimic Bourdain's speech patterns."
No comments:
Post a Comment