Adobe has unveiled “Photoshop for voices”, a tool which lets users edit the content of voiceovers as if it were text.
The software learns to mimic a speaker’s voice, and then seamlessly inserts phrases they never said into the recording, in the sound of their voice.
It could be a boon to the recording industry, but academics have warned that it raises a host of ethical questions.
In a live demo, Adobe shows how a recording which originally said “I kissed my dog and my wife” can be altered to say “I kissed Jordan three times”.
The software requires just 20 minutes of someone’s speech to mimic it, according to Adobe.
But Dr Eddy Borges Rey, a lecturer in media and technology at the University of Stirling told the BBC that the ability to manipulate digital media makes it difficult for lawyers and journalists to use it as evidence.
He added: “In the same way that Adobe’s Photoshop has faced legal backlash after the continued misuse of the application by advertisers, Voco, if released commercially, will follow its predecessor with similar consequences.”
UCL’s Dr Steven Murdoch, a cybersecurity expert, told the BBC the tool also raised security concerns:
“Biometric companies say their products would not be tricked by this, because the things they are looking for are not the same things that humans look for when identifying people.
“But the only way to find out is to test them, and it will be some time before we know the answer.”
An Adobe spokesperson told the BBC the software “may or may not be released as a product or product feature”, saying that “no ship date has been announced.
The spokesperson added that the firm was looking at how to detect use of the software, potentially by using watermarking detection.