Tag Archives: CereProc


Having recently attended Music: An Explanation by a Guitar Hero, which concluded with some deliberations on prosody (the music of speech which amplifies meaning), I chanced upon an inspirational TED talk by film critic, Robert Ebert, who lost his lower jaw, and his speech, through cancer.

Exploring text-to-speech technology, he found that, unless he entered very time-consuming XML coding, the prosody was never quite right. Work is currently in progress with Edinburgh-based company, CereProc to refine his voice, using recorded material from Ebert’s television archive. Exploring their site, I was quite astonished at how far along the speech synthesis road things have travelled. You can hear some of their voices here or type in your own text and choose a voice here. While CereProc finish their refinements, Ebert is using Apple’s Alex voice.

It is very touching to see how Ebert responds during the talk. The words are his own but his wife and two other close friends help out with reading. Despite the fact that the oral delivery is at one remove, he gestures as though delivering the words personally.

Let me, once again, flag up some interesting lectures on prosody by Peter Roach.