Computers: The voice of authority - Tired of a sullen and sometimes reluctant response from his PC, Nigel Willmott tries to improve his productivity by talking to it

Click to follow
The Independent Online
Voice control is not for the socially insecure. If people stare over your shoulder while you are talking to them at parties or sit on you in cinema seats as if you were not there, a rebuff from your computer could be the last straw.

But the Red Dwarf vision of a computer that responds to human commands - at least more often than the available humans - has a great attraction, particularly to keyboard illiterates. The problem is, will it respond reliably. Voice recognition technology is probably at the stage of video processing six or seven years ago. Standard sound cards may allow some basic recognition, but to recognise spoken words reliably, rather more raw processing power is needed.

This means adding an extra 'card' - an additional circuit board - to the computer with a dedicated processor chip. Several small specialist companies like AVT, based in Nottingham and north London, are developing voice systems primarily for business uses in areas like order entry for catalogue sales companies, remote control of cameras in television studios, and medical systems where hand-free control is needed.

However the art has advanced and prices fallen to the point where direct sales to ordinary mortal users is feasible. I first fell for the idea at a computer show, where AVT had cunningly linked the control system up to an ad hoc living room with lamp, fan and television. A quick tutorial and there you were comfortably seated, turning on and off lights and air conditioning, and zapping from channel to channel without all the exhausting effort of pointing and pressing the remote control.

The ultimate couch potato system is still some way off. The main aim of AVT's Voice Server system is to humanise Windows, the PC-compatible's mouse-driven operating system, and Windows programs like word processors.

Installation of Voice Server suggests a little more work is still needed on the consumer interface - though it was probably no worse than many far more established products. It was necessary to switch between two manuals to follow the instructions and though it warned that some small 'jumper' switches on the card might need to be changed to avoid clashing with other circuit boards, like a modem or sound card, it gave no diagram of the switch settings.

Then came the Hamlet cigar moment. The hardware fixed in, the software loaded and ready to go; only the microphone was left to be attached. The basic microphone can be stuck on your keyboard or monitor with velcro, but there can be problems with ambient noise - hair being pulled out and so on. A headset provides much more reliable input, but the cable was so short that when attached to the back of my tower system under the desk I could only use it with my eyes just level with the keyboard.

Using the standard microphone was at least a sterner test of Voice Server's ability to recognise verbal commands. First you have to 'enrol' the words you are going to use - match the words to your pronunciation of them. Voice Server starts with a list of 21 suggested words to teach the computer, the basic commands from the Windows Program Manager. The word required is selected with the mouse and you are then asked to say it three times. If there is too much inconsistency between the versions, it's three times and out - you have to start again.

It was relatively easy to get Voice Server to recognise simple commands like Open and Close, but I could not say a simple phrase like Program Manager consistently three times in a row. It took six attempts before the computer finally deigned an acknowledgement.

The next obvious thing was to move on to a word processor. Unfortunately, Word for Windows version 2, which I would normally use, turned out to be one of the few Windows programs which Voice Server would not work with - it was now beginning to hear a few non-computing terms. Undaunted I tried it with Windows Write.

A first success was teaching it to run the program by voice commands. My 'desktop' - opening Windows screen - is probably not dissimilar from many other people's - chaotic. Trying to find the icon to click on and launch a program can be a couple of minutes dredging through various windows. The spoken Launch command takes you straight to the list of programs and a repeat of the command will load the program.

My test was a memo with lots of numbering and headings, so many changes of typeface from normal roman to bold and italic text. If you are writing, whether it is a letter to the bank manager or the great novel, mixing text input with formatting that text, either using keyboard commands - if you can remember them - or the mouse is a great distraction to the flow of words. And here Voice Server proved a real boon - at least after I had 're-enrolled' the commands Bold and Regular with head down looking at the keyboard, rather than in the carefully enunciated manner I had started with.

Another bonus was being able to deal with stray windows appearing on the screen as a result of accidently hitting control keys with with my dyslexic typing. Instead of having to reach for the mouse, move the pointer to the Cancel box and click it off, a quick verbal 'Cancel' command and it is gone.

The benefit of an effective voice control system is obvious for business uses like order entry, where the operator often repeats the order aloud to check it anyway, or for disabled people with visual or muscular impairment - it can even cope with speech impediments as it only needs consistent sounds to match with the on-screen commands, not Home Counties English. And with the spread of repetitive strain injury in intensive users, anything that cuts down key strokes or mouse movements has added value.

But its benefit for ordinary personal users will depend on how much formatting or moving about Windows the user does. Many people who have to concentrate all their efforts on simply typing will find it a much less distracting way of working. And just as people bond with their pets by talking to them, talking to the computer could make a lot of people feel that it is not quite such an alien lump of metal and silicon.

Voice Server; AVT, 071 454 1224; pounds 292.57 (inc VAT).

IndyBest product reviews are unbiased, independent advice you can trust. On some occasions, we earn revenue if you click the links and buy the products, but we never allow this to bias our coverage. The reviews are compiled through a mix of expert opinion and real-world testing