Last summer, Lexi Bogan, a woman who enjoyed singing Taylor Swift and Zach Bryan songs in the car, lost her voice. Back in high school, she was a soprano in the chorus.
However, everything changed last summer. In August, doctors had to remove a tumor from the back of her brain. After a month, when the breathing tube was taken out, Lexi found it hard to swallow and struggled to even say a simple “hi” to her parents.
Though she went through months of therapy to help her recover, her speech is still not back to normal. It’s difficult for her friends, strangers, and even her own family to understand what she’s trying to say, as reported by ABC News.
Using AI voice replica technology
In April, at the age of 21, Lexi regained a version of her old voice not through natural means but with the help of artificial intelligence (AI).
This AI-generated voice, created from a short recording of her teenage voice, allows her to speak through a smartphone app. The recording was taken from a cooking demonstration video she made for a high school project, capturing her voice at that time.
Now, with just a few taps on her phone, Lexi can have the AI voice say almost anything she wants. For example, during a visit to a Starbucks drive-thru, she used the app to order her drink: “Hi, can I please get a grande iced brown sugar oat milk shaken espresso,” her AI voice said as she held the phone out the car window.
Experts have warned that artificial intelligence voice-cloning technology can amplify scams, disrupt elections, and impersonate people without their consent.
But these Rhode Island doctors believe they’ve found a use that justifies the risks. Meet Lexi Bogan. pic.twitter.com/sWjECORw0R
— The Associated Press (@AP) May 13, 2024
Experts caution that the rapid advancements in AI voice-cloning technology carry significant risks.
This technology has the potential to fuel phone scams, manipulate democratic elections, and infringe upon the dignity of individuals—whether they’re alive or deceased—who never agreed to have their voices replicated to say things they never said.
Instances have already emerged where AI-generated voices were exploited for malicious purposes. For example, there have been reports of deepfake robocalls targeting voters in New Hampshire, imitating the voice of President Joe Biden.
In another case, a high school athletic director in Maryland faced charges for using AI to fabricate a fake audio clip of the school’s principal making racist remarks, according to ABC News.
Justifications that outweigh the risks
Bogan, along with a team of doctors from Rhode Island’s Lifespan hospital group, believes they’ve discovered a justification that outweighs the associated risks. She stands as one of the first individuals to successfully have her lost voice recreated using OpenAI’s Voice Engine.
Similar technology has been tested by other AI providers, such as ElevenLabs, for individuals dealing with speech impediments and loss. For example, a lawyer now utilizes her voice clone in legal proceedings.
Dr. Rohaid Ali, a neurosurgery resident at Brown University’s medical school and Rhode Island Hospital, expressed optimism, stating, “We’re hoping Lexi’s a trailblazer as the technology develops.” He emphasized the potential benefits for millions of people affected by debilitating strokes, throat cancer, or neurodegenerative diseases.