With 2001 looming, you may be starting to look around for HAL -- and wondering why your computer doesn't answer back to you.

Actually, it could, thanks to voice recognition software.

What is voice recognition?
Voice recognition technology allows a computer to recognize your speech and process it as commands or as text to be recorded. You can now find the technology popping up everywhere from your desktop to your PDA to your cell phone.

But don't expect to throw away your keyboard just yet. Although voice recognition is far more accurate than it was at its inception a few years ago, it will probably be some time before you should rush out to suit up every desk in your office with voice recognition software and compatible peripherals.

Still, by now it's certainly worth investigating the field.

The scope of the software.
Probably the most heralded integration of voice and computers is the voice recognition software for your desktop computer.

Some software packages are designed to work with all aspects of your PC, from the Start menu to Internet surfing and word processing. (Expect to pay around $150-$200 for these packages).

Other packages are more specialized, geared towards spreadsheets or composing e-mail, for example. Their limited capabilities are reflected by lower prices: around $50 or less.

How accurate is the technology?
In general, voice recognition is not accurate enough to be useful for long dictations, but for a short e-mail it could be just fine.

Accuracy for most packages tends to hover at the 90% mark for interpreting rather natural speech, meaning that the computer misinterprets one word out of ten.

The best packages are still only at 95% accuracy, or one word in twenty.

While it may seem like a no-brainer to just choose the program with the highest level of accuracy, it's not that simple. For one thing, accuracy depends on compatibility between software and hardware (sound cards, microphones), as well as the ability of the software to learn YOUR speech -- particularly if it's laced with jargon.

Learning to work with the software.
All programs come with a standard vocabulary, but fixing mistakes is what takes time. Impressive accuracy initially might not seem so wonderful if correcting that irksome 10% is an ordeal.

Voice recognition is not perfect technology; it takes some time getting used to. There's a period of "enrollment" where you train the computer to understand how you speak (about 10 minutes).

Then, getting fully accustomed to the software could in fact take months - but if you're patient and buy the right package, it can work for you.

Choosing a package.
When choosing between packages, you'll find the difference in the details.Try to talk to software users (try on-line forums) about their experience or grill service reps for specific questions.

• Is dictation and using commands seamless or does it make you stop speaking?
• Will it recognize proper names and places?
• How does capitalization work?

If voice recognition is something that would make your life easier, it's certainly worth checking into and getting used to. Then all you'll have to deal with is explaining to people why you're talking to your computer.

Quick tips