"Have you heard Roger Ebert's new voice?" he asked. I hadn't, but knew that the film critic/commentator was unable to speak, owing to the terrible effects of cancer. "Check it out, it's amazing," he said. "They used old clips from his TV shows to build a library for computerized speech."
Intrigued, I did. A demo featuring Roger and his wife, originally played on the Oprah Winfrey show, shows what it's like:
Note there are two synthesized voices he uses -- a synthetic male voice from a standard voice software program, and a new product developed by a Scottish company, CereProc. For decades, until the effects of cancer killed his voice, Roger co-hosted a popular television show that covered new film releases. Through the program, he developed a wonderfully smooth talking and debating voice (he frequently sparred and joked with his first co-host, the late Gene Siskel). CereProc was able to leverage this library of sound, so when Roger types a sentence on his laptop, the program "reads" out what he said, stitching together words from his old television programs.
Is the new speech technology from CereProc perfect? No. But it's much better than the synthetic male voice, and really shows where the technology is heading. I can see people in the future "training" a synthesizer using their own voice (or old audio/video clips) for all sorts of functions. It will be a godsend for people who have lost their ability to speak, but can also be applied to services ranging from voice mail/IVR, news programs based on the "voices" of famous (or obscure) people, and machinima.
More blogging about virtual worlds:
- Hug a Pug: A virtual world lesson for the real world music industry
- The Tiger Woods accident simulated in 3D shows the future of newscasts
- The open-ended first-person shooter: A template for 21st century storytelling?
- Microsoft's virtual worlds strategy: It's the 'Spatial Web', not Second Life
- My new media manifesto: "Meeting the Second Wave"