I kid you not. Hot on the terrifying heels of abandoned satchels, station bins and flaming cars comes the latest thing to be afraid, be very afraid of: computerised speech. Synthetic voices capable of mimicking human voices could create a form of "vocal terrorism", British scientists this week warned.

And it's not (just) by mimicking the Daleks or creeping up behind you and shouting real loud, it's even scarier. Apparently, computerised voices are becoming so good, they could be used to spread disinformation via telephone networks. (They may even get good enough to tell you how delayed your train will be without sounding like a dog with 17 gobstoppers on the go).

Scientists believe that by recreating digital versions of people's vocal tracts, synthetic voices will become indistinguishable from real human speech in the next 15 years.

York University's David Howard produced 3D computer models of a man's larynx using medical scans of the throat, supplied by Brad Story from Arizona University. By manipulating the virtual larynx, the pair recreated the movements of the throat used to produce different sounds.

Dr Howard warned that, while convincing synthetic voices will improve some aspects of society, they could create new forms of crime and disruption.

"It could be that you're fooled into thinking your bank manager is ringing you, and gets you to divulge details of your account, or it could be that I've taken over a communications network for a country and I broadcast the sound of a leader to the people," he added.

Which leaves one question. Why do it?