CHICAGO (Reuters) - Film critic Roger Ebert, rendered mute by several cancer surgeries, delivered his Oscar picks on Oprah Winfrey’s television chat show on Tuesday using his newly synthesized voice fashioned from old audio clips.
“I can’t remember a year when it was easier to pick the Oscars. Those may be famous last words,” the now cancer-free Ebert, 67, said in a recorded segment on “The Oprah Winfrey Show.”
“The only dicey category is Best Picture,” said Ebert, who copyrighted his famous “thumbs-up” or “thumbs-down” ratings created for his pioneering televised review show with Chicago Tribune critic Gene Siskel, who died of cancer in 1999.
Ebert predicted Kathryn Bigelow’s “The Hurt Locker” would win the Oscar for best picture, and Jeff Bridges and Sandra Bullock would win for best actor and actress, respectively.
Ebert won a Pulitzer Prize in 1975, the first film critic to do so. Since being diagnosed with thyroid cancer in 2002 he has undergone numerous surgeries that have left him without most of his lower jaw and cost him his voice.
But he communicates by writing notes, uses crude sign language and “speaks” with the aid of a software program loaded onto his laptop computer that generates a generic voice.
A Scottish company, CereProc, that was spun off five years ago by University of Edinburgh researchers, is developing a program just for Ebert that culled hours of his old movie commentaries to give him back a semblance of his own voice.
“In my dreams, I‘m talking all the time,” Ebert told Winfrey. “That’s just like I was in life. You could not shut me up.”
Ebert, who remains a prolific film reviewer for the Chicago Sun-Times and sees as many as four movies a day, called his new voice “Roger Jr” in a recent column for the newspaper.
“Yes, Roger Jr needs to be smoother in tone and steadier in pacing, but the little rascal is good. To hear him coming from my own computer made me ridiculously happy,” Ebert wrote.
Winfrey’s cameras captured the tearful reaction of Ebert’s wife Chaz hearing his voice for the first time in years.
“What do you think?” she asked Ebert. “Uncanny. A good feeling,” he tapped out on his laptop in reply.
CereProc’s technology officer Matthew Aylett said in a telephone interview the goal is to produce voices that provide “personality and depth of character.” The technology has yet to advance to the point where a person could replicate shouting.
Ebert’s synthesized voice sounded somewhat flat and lacked inflection.
“This is the first version of my computer voice. It needs improvement but it sounds like me,” Ebert told Winfrey.
The cost of CereProc’s software with generic voices -- such as a British-accented voice dubbed “Lawrence” -- is roughly $30 to $40, Aylett said. He would not estimate the cost of a customized voice.
In the future, synthesized voices might add warmth to interactions with robots or virtual reality avatars, Aylett said. Someone about to undergo surgery could record their voice beforehand and have it reconstructed after, he said.
Editing by Eric Walsh