Term explained
What is AI voice cloning?
Voice cloning is using AI to copy a real person's voice from a recording, so that it can say any words typed in, sounding just like them.
In plain English
Voice cloning tools listen to a recording of someone speaking, learn how their voice sounds (its tone, accent and rhythm), and then read out any new text in that voice. Some tools need several minutes of clean audio; others manage a rough copy from a few seconds, such as a clip from a social media video or a voicemail greeting.
It is the audio half of a deepfake, and it is often more convincing than fake video, because a phone call gives you fewer clues to check.
Why it matters
Voice cloning has real benefits. It can give people who have lost their voice to illness a way to speak in their own voice again, and it lets creators dub videos into other languages. But it has also made an old scam much more convincing: the call from a “relative” in trouble who urgently needs money, now in a voice that sounds exactly like them.
An example
A parent gets a call from what sounds exactly like their son, crying and saying he has been in an accident and needs money for a lawyer right away. A “lawyer” then comes on the line with payment instructions. The son is in fact safe at work; scammers cloned his voice from videos he had posted online.
What to watch out for
Agree a family code word that a real caller in trouble would know. If a call is urgent and about money, hang up and ring the person back on the number you already have. And be aware that “my voice is my password” checks at some banks can be fooled; use stronger security where you can.