Computers could mimic human speech so perfectly that vocal terrorism could be a new threat in 10-15 years' time, scientists suggest.In the future, it may be possible to mimic someone's voice exactly after recording just one sentence. Such technologies would pose a danger if it were not possible to verify who was speaking, researchers believe.
New methods of creating computerised speech use models of a vocal tract to create a realistic sound, replacing the existing technique of copying sounds.
"We are beginning to simulate virtual vocal tracts in the computer," said Dr Howard. "When we simulate this in the computer, which we are beginning to do, we begin to get sounds that musicians describe as organic or more natural.
"If we get to the point where we are synthesising the actual shape of somebody('s vocal tract) based on analysis of their speech, then the speech we are producing should sound and look like the actual person."
Source: BBC
Tags: Speech | Vocal
Comments