Communications of the ACM
How Google Plans to Clone You
Google recently acquired Phonetic Arts, a U.K.-based company specializing in technology that transforms a recorded voice into a computer-generated voice that sounds like the recording. In other words, it "captures" the tonal qualities, cadence and rhythm of how a real individual person talks, and applies them to a machine voice. The result is that a computer will be able to read any text, and it will sound convincingly like the original speaker talking. The voice Google wants to capture is yours.
Better than Star Trek
In Gene Roddenberry's original "Star Trek" TV series, the characters interacted with computers by talking, and the computers talked back. Although this was a breathtakingly advanced concept in the late 1960s, it turns out that the real future is far more interesting.
In 1960s sci-fi, voice interaction with a computer was generic—used for input and output, for commands and responses. It wasn't customized, and it certainly wasn't personalized.
A much more accurate fictional account of where voice interaction is going comes from William Gibson's 1984 novel Neuromancer. In that book, people have virtual versions of themselves, represented by a 3D computer scan of the person's face and a computer-generated version of their voice, backed by artificial intelligence and data about the real person.
Gibson expanded on the concept in another novel, called Mona Lisa Overdrive. In that work, people could record their personalities on storage media: "They respond, when questioned, in a manner approximating the response of the subject."
Google's vision is more Gibson than Roddenberry.
From Computerworld
View Full Article
No entries found