What is a Voices Cloning?
Cloning voices is creating an artificial simulation of a person’s voice. Methods of today’s AI voice cloning software are able to generate synthetic speech that closely resembles a targeted human voice. In some cases the average person is unaware of the difference between the real and the fake voice.
Sounds like a deep-faced voice clone used in a failed fraud attempt.
The audio deep fake fraud is one of the stranger uses of deep fakes — AI technologies used to exploit audiovisual information. Hackers use artificial intelligence to replicate someone’s speech and then pair the speech replica with tactics of social psychology to induce people to transfer money where it shouldn’t. In the past, these hacks were effective, but how reliable are the voice clones used in those attacks? Unfortunately, we have never detected the audio from a deep fraud — so far.
NISOS also published a study investigating one such alleged scam, and shared the audio with Motherboard. The snippet below is part of a voicemail sent to an employee of an unidentified software corporation, in which a voice that sounds like the CEO of the organization requests the employee for “immediate assistance in finalizing an urgent business contract.”
The quality is certainly not great. Even under the cover of a bad phone signal, the voice is a little robotic. But that’s only passable. Even if you were a novice employee, concerned after your supervisor gets an apparently urgent call, you would not worry too much about audio content.”Sure it sounds normal. They checked the box so far as: does it sound more human or more robotic? I’d say more human,’ Motherboard told Rob Volcker, a NISOS researcher. “But it doesn’t necessarily feel good enough for the CEO”.
In the end, the attack was unsuccessful, as the employee who received the voicemail “immediately thought it suspicious” and flagged it to the legal department of the firm. However these attacks should become more popular as deep-fake techniques become even more available.
All you need to build a voice clone is access to lots of target recordings. The more data you have, and the higher the audio quality, the better the resulting voice clone. And those recordings can be easily collected from earnings calls, interviews, and speeches for many executives at large firms. The best quality audio deep fakes are far more compelling than the illustration above, given enough time and details.
The best-known and first-reported example of a deep-fake audio scam took place in 2019, where a UK energy company’s chief executive was tricked into sending €220,000 ($240,000) to a Hungarian supplier after receiving a phone call allegedly from the CEO of his parent company in Germany. The executive was told the transfer was urgent and the funds had to be sent in an hour’s time. He did. The perpetrators never got captured.
The FTC cautioned about the emergence of these schemes earlier this year but experts believe there is an simple way to combat them. As the Herbert Weathermen College of Engineering’s Patrick Tray nor told The Verge in January, all you need to do is hang up your phone and call the person back. For other scams the attackers use a burner VOIP account to reach their targets, like the one mentioned by NISOS.
“Hold on and call them out,” Tray nor notes. “Without becoming a state agent capable of rerouting phone calls or a very, very advanced hacker squad, odds are that’s the easiest way to find out whether you spoke to someone you figured you were.”
Kevin Smith is a Microsoft Office Product Technical Assistant for 10 years who guides the user to setup an office product and reviews many websites which are office.com/setup and www.office.com/setup.