The efforts to make a computer speak as we do or at least type for us are making progress. It’s good progress, too. You can now chat away to your word processor without being carted off to a padded room. And the days of culturing your flattest voice when using speech recognition software are over - if you speak in a clear voice, the keyboard can be left alone.

However, a level of expectation must be set. Speech recognition is best done on the machine you use all the time and is better still when you’re proficient. Like learning to touch-type, using software needs persistence and like any situation with learners, those offered a taste of salvation will need encouragement to get over the initial blues.

People say things like, “Speech recognition? Yeah, it really works.” And they’re right - sort of. Tell the machine that and it might type “Each patient care they were,” instead. Repeat the test to the point of frenzy and the result may be no worse: sense cannot be had from the thing. Yesterday it seemed intelligent, now the microphone volume is awry, the software is sick and you’re shouting.

There are two types of software. One is the discrete speech recognition package that requires less powerful computers and processes individual words; many schools with younger children say this is best. The other allow continuous speech, which is fast but requires a fast machine. This software is more common, with updated versions of packages like Voice Xpress (see right) and ViaVoice (opposite) appearing annually.

After that you need the right gear - and the right gear is not always the latest. Always check advice that comes with the software; the microphone supplied may work, but it may be too flimsy for school use. Look at the small print and it says a USB-type microphone gets around a problem sound card and that laptop computers often aren’t up to the job (often because of compromises with the sound card). Further advice points out that you’ll need a computer memory of, say, 128Mb. Don’t pay attention to these points and you may experience software failure or dreadful frustrations, for which the publisher cannot be blamed.

Then it’s time to do some reading, but the text is too tricky for some children. Enter the British Education and Communications Technology Agency. It has worked with a number of schools to analyse this technology and offers user reports and advice through its website at www.becta.org.uk.

Finally, you read the machine a bedtime story so it can tune itself to the sound of your voice. And assuming you don’t throw it out of the window in sheer frustration, you and it will eventually learn to understand each other. You will bond and get some work done. Now, time for my medicineI If an early taste of speech recognitio software raised an eyebrow, a serious look at version 5 of Lernout amp; Hauspie’s Voice Xpress might make an eyeball pop.

Working at speeds of 130 words per minute - quicker than the fastest typists - this is seductive technology to watch in action. Unlike discrete dictation packages that process single words, Voice Xpress uses the probability of one word following another to generate impressive results. Its ability to “think” in context lets users talk naturally, pausing only to say “comma” and “full stop”.

Use Voice Xpress and the fantasy of dictating a school report to the computer as you put your feet up seems almost realisable. You talk as fast and as clearly as you can while the machine types out the words in swathes. The reality is you correct errors as you go but you can do this entirely hands-free. For example, you’ll say “correct ‘realise of all’” and spell out its correction (“realisable”). Dialogues may appear with alternatives like “to” or “two” and you say which you want. And you train the software to learn specialist vocabulary or acronyms - ICT need no longer be returned as “icy tea”.

There’s clever technology behind this. It surprises by making sense of phrases like “you need a 200MHz Pentium”, but it can be a bit too clever and interpret “from five to ten” as “from 9.55”. Perseverance teaches you its quirks and there’s a way of speaking that works and others that do not.

The problems I had were outside the realm of reasonable complaint: I could not think fast enough for a machine that prefers clauses and it didn’t like my tongue slips, dry lips or addiction to chewing gum.

First you must train the package to recognise your voice. This takes under 10 minutes and creates a profile you can keep if you change machines. As you’d expect, transcription happens in any application you use; for example, you can talk straight into Microsoft Word, Excel or even Windows Notepad.

The package comes in various flavours, for example offering legal or medical words or some offering extra voice commands listed on-screen, but less may be best. You can control most of the varieties or say “bold that; make it red; bullet that”. When used to transcribe tape recordings (by voicing the words into the microphone) a mixture of talking and mouse clicking is more efficient; nice as it is to work with feet up, it’s easier to click things than mouth words.

Voice Xpress Standard

Version 5 of Voice Xpress will filter out ums and ers and provide more voice control but best is RealSpeak, which reads words on screen in a natural voice. This may well be the sound of tomorrow’s bedtime story.

Voice Xpress Standard Price: pound;39.99

Voice Xpress Advanced Price: pound;79.99

Voice Xpress Professional Price: pound;119.99 All come with headsets and require a minimum Pentium II processor. Check with Lamp;H Tel: 0800 056 0539 www.lhsl.com

ONLINE STAR RATING

Suitability for purpose: ***

Ease of use: ****

Value for money: *****