Back when Captain Kirk barked his orders at the helm of Starship Enterprise and actually got a response, we all knew it was about as phoney as the set which surrounded him. But while the command to "Beam me up Scotty" may still be impossible, voice-responsive software is now a reality. And it is affordable.
Several companies specialising in voice-recognition technology are gearing up for a market offensive over the next few months with products which bring the application to a new, user-friendly level.
Voice computing enables users to control their computers through speech. The implications for personal and business computing environments are immense, particularly with the growing availability of continuous speech products which no longer require users to speak in a robotic fashion - pausing between words - as was the case with earlier software.
Belgian company, Lernout & Hauspie (L&H), has just released a speech-recognition product, Voice Commands, which allows the user to format and edit text on Microsoft Word 95 or Word 97 without having to learn to use Word. It carries out the functions purely through voice commands.
A document may be formatted by saying: "Go to the beginning of the document. Select the first paragraph. Centre it. Make it 10 points bigger." L&H which has an operation in Blackrock, Dublin, currently leads the field in speech technology innovation. This was copperfastened recently when Microsoft invested $45 million (£30 million) in its speech products.. The company's revenues jumped from $17 million worldwide in 1996 to $65 million by the end of September. Its clients include high-profile operators, including Northern Telecom, Ford and Samsung.
The Voice Commands product was developed on foot of a general recognition by L&H, that while people can often type quickly, the tidying up of documents takes time. One verbal command can now replace 10 mouse clicks.
Voice Commands, which is expected to be in the shops by Christmas, works independently or in conjunction with L&H's dictation package, Voice Pro, which retails at £149.95. With a 60,000 word vocabulary, users can dictate documents straight onto the screen.
"Because of the significant advances made in the area of voice technology in the last year, what was a dream two years ago, is an affordable reality now," says Mr Mark Kent, sales director of L&H in Dublin.
US Software group Dragon Systems is rising to the challenge by offering NaturallySpeaking dictation software in a new "Preferred Edition" that contains most of the features found in its Deluxe edition but costing significantly less at $199 compared to $695.
Dragon Systems says it launched the first general purpose, large vocabulary, continuous speech recognition product last April. But L&H was not far behind with its acquisition of the US software group, Kurzweil, for $53 million. The move is thought to have enabled L&H to bring its continuous speech product to market six months earlier.
"By the year 2000 the keyboard will have become like the mouse - handy to have if you want it, but the voice will take over. Now people can use free speech, there is no limitation to the potential," says Dr Peter Kelway, managing director of Aptech UK, supplier of state-of-the-art speech technology hardware and software.
While these products are targeting the mass market, particularly small or home offices, the advantages for people with physical disabilities are very significant.
Major player IBM - which employs more than 500 people in Ireland - has also been involved in the race, developing viaVoice Gold, which is to be launched in the US within two weeks and in Europe in the new year. Expected to retail in Britain at £139 sterling, viaVoice Gold is the latest version of viaVoice System, and can run on any desktop programme. It can handle more than 100 words per minute with 95 per cent accuracy.
But not everybody is convinced voice-recognition technology is a time-saving device just yet. "The products really need to be perfect before they can be adopted by the general market. Accuracy rates of 95 per cent sound very good, but when you consider that out of 1,000 words there will be 50 errors, that can be very time consuming," says Prof John Byrne, head of Computer Science at Trinity College, Dublin.
Only a decade ago, speech recognition demanded the power of a massive mainframe. Since then, advances have been accelerated by Intel's Pentium processors which provide the desktop processing power.
