How the world's most famous film critic got his real voice back

Cancer left Roger Ebert reliant on an artificial voice machine. But now, by using recordings of his old broadcasts, it sounds just like him

Four years ago, Roger Ebert, America's most famous television film critic, thought he would never appear on screen again. The 67-year-old reviewer has been a household name in the US since launching his TV show, At The Movies, in 1982. But he had to give up live broadcasting in 2006 when he lost his voice after operations for thyroid cancer, in which his lower jaw was progressively removed.

Now he will appear on screen for the first time since the surgery, on The Oprah Winfrey Show, using his own voice. For the past four years, Ebert has communicated through hand-written notes, simple sign language and with the aid of a basic voice synthesiser, similar to the one made famous by Stephen Hawking.

Then last summer, while trawling the internet, he came across a Scottish company, based at the University of Edinburgh which produces customised voice synthesisers that translate text to speech and deliver it in the voice of its author.

Ebert was impressed by what he heard – samples on the website include a synthesised Barack Obama talking slightly more hesitantly than the real thing, and Arnold Schwarznegger. They were a world away from the monotone Hawking has been famed for since in 1980s. Encouraged, Ebert contacted the Scottish company, CereProc. He wrote at the time: "I have my fingers crossed. I can see my own voice hosting online or telecast video essays. I am greatly cheered."

This week he took delivery of a prototype and is said to be pleased with the results and keen to demonstrate it in public.

Fortunately for Ebert, his archive of TV tapes, DVD commentaries and radio shows provided a rich seam for the company's researchers to mine. They painstakingly reconstructed his voice, piecing together snippets of recordings to produce new words and sentences. Ebert now has a database of words and a system which can "speak" any typed sentence in the voice he had before his operation.

It will not sound exactly as he used to but it will sound more like him than his existing generic voice does. In time, he should be able to modulate the new synthesised version to make it sound more natural.

Matthew Aylett, the chief technical officer of CereProc, said: "One of the things that we specialise in is trying to produce voices which have got a bit of character and don't sound neutral or boring. In this case, we're using audio that has been recorded for commentaries on Casablanca or Citizen Kane, for example. We have to take this audio and try to produce something which sounds smooth and natural." It is believed to be the first time that archive audio material has been used to benefit the speaker.

Mr Aylett said: "We're giving Roger Ebert actually the same voice he had before he had surgery. When he uses it, people who listened to his commentaries in the past as a broadcaster will recognise his voice. Both are examples of the way the voice is so central to who we are as people and so much of our character and personality is expressed by the use of our voice."

Ebert, the first film critic to win a Pulitzer, in 1975, hopes to produce an audio version of his popular blog and is wondering whether he can use it on radio and television.

Hear CereProc's voices at http://www.cereproc.com/products/voices