Voice authentication has become a standard in recent years but now it appears software has bitten with a Canadian startup, Lyrebird, demonstrating how they can mimic people’s speech.
Last year at one of their industry events Adobe showed off their ‘photoshop for voice‘ where anyone’s voice can be analysed and then remixed.
So voice recognition turns out not to be a foolproof as many in the security industry hoped. Like most biometric systems, anything that can be captured electronically can be spoofed or modified.
What’s notable in the Lyrebird story is how voice security companies like Nuance are deploying artificial intelligence to counter digital fakes. Once again we see security being a technological race.