News :: [none]
MIT develops fabricated lip-syncing video technique
||| COSMIVERSE - Scientists at the Massachusetts
Institute of
Technology
are scarily good at putting words in your mouth. . .
They've developed
technology that can analyze a videotape of someone
speaking, "learn"
what their mouth looks like when forming certain
sounds, and then
literally string those sounds together in order to
form just about any
words you would like the image to say.
Even scarier,
they're able to
fool the average viewer on a consistent basis, so much
so that people
are unable to distinguish between the actual person
speaking and the
animated "copy" speaking. In one demonstration, they
were able to take
footage of a woman speaking and had her lip-synching
to a song in
Japanese, a language she does not even speak. . . The
opportunities for
misusing the technology are almost as limitless as the
potential it
holds to enhance the entertainment and special effects
industries
www.cosmiverse.com/news/beaten/beaten05170202.html