The cause of life-like Text-to-Speech rendering is about to take a giant step forward when film critic Roger Ebert appears on Oprah. For those who have not followed Mr. Ebert’s personal history, he has spent the past few years unable to eat or speak after undergoing cancer-related surgery. , excerpted from an article he wrote for Esquire magazine is one of the most moving personal diaries I have ever read, and provides testimony to the quality of his mind and spirit.

, Ebert describes how the discovery of a Scottish company called CereProc convinced him (and his wife) that the automated system could render his voice with the pronunciation, inflections and levels of enthusiasm characteristic of his banter in better days. As Ebert explains, his wife Chaz “could tell it was me. For one thing it knew exactly how I said ‘I.'” And, as Ebert himself wrote about hearing the rendering, “To hear him coming from my own computer made me ridiculously happy.”

And now, for the rest of the story. The Edinburgh address was something of a tip-off. The company’s chief technology officer, Dr Matthew Aylett, was the senior development engineer at Rhetorical Systems (which was acquired by Nuance – during its days as Scansoft in 2002). In that role he “was responsible for the design, implementation, and testing of the company’s core speech technology” which differentiated itself from the pack with its “lifelike” rendering of text. At CereProc, he’s joined by Christopher Pidcock, who is described as “the architect of Rhetorical’s state of the art voice creation system.”

Just as Oprah and Ashton Kutcher fueled a media-driven craze that led to a meteoric rise in awareness of Twitter, the popular TV host, with a special assist from Roger Ebert stand a good chance of popularizing life-like TTS. It is fitting that Rhetorical Systems’ alums stand to benefit, but other technology providers – including Nuance, Loquendo, SVOX, Wizzard Software, Acapela, among others- will benefit as well.

Categories: Articles

Tags: Recombinant Communications, Text-to-Speech, Voice processing