Voice of reason

22 October 2009 8:00 am

Science fiction has an uncanny knack of predicting the future. Over the last 20 years, many things that would previously have been thought of as fantasy have now become fact. Mobile communication devices, handheld personal computers and electronic books all featured prominently in the original 1960s Star Trek series yet today these devices are increasingly part of our everyday lives.

Another part of Gene Roddenberry’s vision of the future was computers that you could talk to. This too is now increasingly becoming a reality.
Although today’s children use technology from the youngest age, many IFAs were educated in an era where typing was a skill only taught to girls. Consequently, they struggle with a computer keyboard.

One technology which has long fascinated me is voice recognition software. It was the subject of one of the very first technology columns I ever wrote 15 years ago and I have revisited the subject periodically every few years.

Until last year, it had been my experience that, each time I tried to learn to use a voice recognition system, out of frustration, my typing ended up improving.

To make voice recognition a practical tool, I have found you need recognition rates in excess of 99 per cent. Having had many false dawns where such systems, while initially promising, were in practice too unwieldy, I promised myself some years ago that I would not write about the subject again until I came across a product that was truly practical to use on a day-to-day basis.

Today I am pleased to report that this has been my experience with the current, version 10, of the Dragon Naturally Speaking product. For the last 12 months, I have been using this package on a regular basis and now find it far easier to dictate emails rather than type them. Having had the opportunity to test the software on a range of different computers, desktops and laptops using different operating systems, I have repeatedly been pleasantly surprised at the reliability of this product.

Historically, it was necessary to spend many hours training the software, reading aloud long passages from books and manually correcting errors when they occurred if you wanted to achieve a high level of recognition.

It has now been my extended experience that with very little training the software can be ready for everyday use.

Using voice recognition software does require changes to the way you work. First, it is still not good if you have a noisy background. This is because human hearing is surprisingly adept at screening out different type of background noise and distinguishing speech from other sounds. At the moment, computers are far less able to make this differentiation.

One of the other lessons I have learned is that you need to get into the discipline of using the system all the time, even for the briefest of one-line emails

It is also important to recognise that the types of error one will make using voice recognition software are different from typical typing errors. For example, spell- checking systems cannot help you if the software has misinterpreted the word you are trying to use. Spelling mistakes are replaced by “sounds like” errors. Thoroughly proofing your communications before you send them becomes essential.

One of the other lessons I have learned over the last year is that you need to get into the discipline of using the system all the time, even for the briefest of one-line emails.

Opening the software and putting on a headset only takes a few moments and while a single one-line email might be completed more quickly by typing, as soon as you begin the next mail in a single session you are beginning to get a real productivity benefit. Most people can speak around three times faster than even an experienced typist can use a keyboard, so it does not take long to get real gains.

At the risk of stating the obvious, it is also important to remember to turn off the software when breaking off for a conversation as otherwise your dialogue can find its way into the middle of the mail you were dictating.

Another historic frustration with these systems was that they tended not to work well with portable computers.

My current Hewlett-Packard tablet PC is without doubt the worst laptop I have ever had the misfortune of working with. However, even on this, the voice recognition is entirely usable.

It is important to invest in a high-quality microphone. But whereas in the past these might have cost a couple of hundred pounds, an excellent portable USB mike can now be acquired for under £30.

There are still a few frustrations with the product, for example, any words which might also be used to navigate around your computer can be used as commands, that is, to tell the computer to do something. In my experience, this means it can be quicker to type these rather than have the software interpret text as a command.

To help with proofing, you can use the included text to speech reader. This reads your text back to you. How-ever, the voice is very synthetic and a little annoying.

A range of different flavours of the Naturally Speaking product are available from around £100 upwards, depending on what accessories you want to come in the box.

It goes without saying that this article was created using Dragon Naturally Speaking and in the process there were only six words that the computer misunderstood.

Having been a long-term cynic about voice recognition, I am now very much in the convert camp. For anyone who cannot touch type and even for many who can, buying a copy of this software could be a very worthwhile investment.

If you enjoyed this article, sign up here to receive daily email updates from Money Marketing and Follow @_moneymarketing