Kleppy writes: "CereProc, an advanced voice synthesis company in Edinburgh, creates customized text-to-speech software. Instead of producing flat computerized voices, the company says its voices include realistic, animated and emotional dimensions. "
The company spliced together a few hours of dialog from prior recordings of Ebert and broke it all down to about 45 main "sounds" he used for speech. They then used these sounds in a text-to-speech software and were able to produce a prototype that he demonstrated on Oprah. They hope to have a fully functional version with complete intonations in the coming months.
Maybe M$ could use this to get rid of the robotic demon they currently use and replace it with Bill himself?