On Tuesday I plan to sit down and watch The Oprah Winfrey Show for the first time, ever. I've seen bits and pieces before, but this will be the first time I watch the whole thing in a premeditated fashion. Hey, I want to hear Roger Ebert's new/old voice.
Ebert's new voice has been synthesized (and is being further refined) from DVD commentary tracks he recorded for a handful of movies. The Scottish company behind the voice is CereProc, which specializes in text-to-speech synthesizers that speak in a variety of accents. It's fun to play around with their live demo and make voices from all around the British Isles say vulgar and juvenile things.
As more and more of us litter the intertubes with extensive examples of our speaking voices, the easier it will be for convincing artificial versions of our voices to be cobbled together. I suppose the technology will have matured when it can pass a sort of text-to-speech Turing test—when someone can call your close friends or relatives by telephone or Skype or whatever and fool them into thinking they're talking to you.
Damn, I just got an idea for a story.
