Voice recognition made easy

There are plenty of reasons you might want to control a PC using speech recognition instead of a keyboard and mouse. Providing you have Office 2002 or a later version, there’s little to stop you trying it out. All you need is a microphone and some time to train the tool to interpret your voice.

For those who suffer from repetitive strain injury (RSI) or have limited manual dexterity, speech recognition may be the only way to control a computer or enter text.

Dictating text can also be much easier than entering it on a keyboard for anyone who is not the speediest typist; conversing in real time with other chatroom users or an instant messaging buddy, for example, can be stilted when typing with only one or two fingers.

Speech recognition can also be extremely helpful for people with dyslexia who can sometimes find typing difficult.

The Office speech-recognition tool is not designed for completely hands-free operation, however. For the best results, you’ll need to use a combination of the keyboard or mouse and your voice.

Speech-recognition technology has had a reputation in the past for patchy performance. Early speech-recognition software was temperamental. It didn’t respond well to regional accents or the kind of speech variations that pepper normal human conversation. The occasional ‘um’ or ‘er’ would throw a spanner in the works and the speed at which the user could dictate was painfully slow.

Things have moved on significantly in recent years. Modern speech-recognition programs such as Dragon Naturally Speaking 9 claim a 99 per cent accuracy rate in converting speech to text. You may not experience this level of accuracy on the first occasion, but the program can be trained to recognise your voice better if you’re prepared to put in the effort.

One vital piece of equipment is needed before experimenting with the speech-recognition tool in Office: a microphone.

Speech-recognition technology works best with the clearest speech input possible so, in theory, the better the quality of the microphone, the better the results. If you already have a microphone at home, try it out and see what sort of results it produces with the speech-recognition tools we’re going to look at here.

We tried out the speech-recognition tool in Office using a basic in-ear headset – the kind used for making internet telephone calls – with an inline microphone built into the lead. It worked well and costs only about £5. A headset is ideal as it can be adjusted and avoids having to lean forward to speak into a microphone on the desk, for instance.

The microphone should be placed about an inch to one side of the mouth, not directly in front where it may pick up breathing sounds that will make it harder for the computer to interpret what is being said. If you’re using a notebook PC there may be a microphone built into the screen surround, but this may not be practical to use for long periods of time.

Before doing anything else, make sure the microphone is properly connected to the computer. If using a straightforward microphone, plug the connector into the microphone socket on the computer. This looks just like a headphone socket and should be labelled with a microphone icon. On most PCs, the socket has a pink surround.

If you are using a headset, there will be two connectors at the end of the lead: one to connect the microphone and one for the earpiece. Plug the headphone connector into the green headphone socket on the computer.

If the computer has a soundcard, ignore any headphone or microphone sockets on the front of the system case. An easy way to check whether or not a computer has a soundcard is to look at the back of the system case. If there are headphone and microphone sockets, then it does. Use these to connect the headset or microphone.

Also, take a moment to make sure that the sound level for the microphone is turned up. If it is not, the computer won’t pick up the dictation.

To turn up the sound, go to the Start menu and click on Control Panel. Double-click on the Sounds & Audio Devices icon, then select the Voice tab in the Sounds and Audio Devices Properties dialogue box.

Under the Voice Recording heading, click on the Volume button, and the Recording Control dialogue box will appear onscreen. Use the slider under the Microphone to set the volume to maximum and ensure the box under this labelled Select is checked. Close this dialogue box and click on the Apply then OK buttons.

Speech recognition can be used in any Office application and is accessible in each program through the Tools menu (in Excel choose Speech then Speech Recognition).

This launches the Language taskbar. Setting it up for the first time, though, has to be done in Word, and as it’s a handy tool for dictating long tracts of text, we’re going to focus on Word in this feature.

To get started, open Word and select Speech from the Tools menu then click on Yes when asked if you want to install the feature.

The next step is to complete a training session to help the speech-recognition tool recognise your voice, which will take about 15 minutes. Click on Next in the dialogue box that appears on screen and follow the instructions provided.

If for some reason there isn’t time to train the speech-recognition tool immediately after installation, you can always return to this point by selecting Tools from the Language taskbar and clicking on Training.

Speech recognition of any kind works best in a quiet room, so if there’s a chance of being disturbed during the training session, come back to it later when the room is quieter.

Those who use a notebook rather than a desktop will need to repeat the training sessions to get the best results if they move the computer to a different location.