Dubbed Dasher, the text entry system could transform computing for people unable to use a normal keyboard.
Designed by David MacKay and David Ward in the University's Department of Physics, the system uses an eyetracker which can deduce where the user is looking on the screen.
Letters continuously appear on the screen, with Dasher displaying the most likely pairings or triplets of letters. It can even guess whole words.
Few spelling mistakes
Dasher's knowledge of English is cumulative so it will remember new words and has already digested some classic authors such as Jane Austen and Lewis Caroll in its attempt to help users write quickly and accurately.
"The software works like a video game in which the user steers ever deeper into an enormous library," explained Dr MacKay.
The system is faster than other eyetracking systems producing up to 25 words per minute compared to the 15 word per minute currently possible.
"Not only is this faster than any alternative writing system driven by an eyetracker, the frequency of spelling mistakes is about five times smaller and the new system is also less stressful to use," added Dr MacKay.
Its inventors stress that Dasher has not caused any eyestrain in experiments.
Useful in China
But users preferring not to rely on their eyes can also use a mouse, rollerball or touchpad.
Dasher will be developed as an open source software project, Dr MacKay said.
It is hoped that the software will be useful for handheld computers or mobile phones where there is no space for a keyboard.
It could also prove invaluable in the burgeoning markets in Japan and China.
Both languages have thousands of characters meaning they are poorly suited to conventional keyboards.
The research was published in the journal Nature.