Windows Speech Recognition

In my previous post I provided a list of more advanced keyboard shortcuts to minimise use of the mouse. The next step then, for the really determined, is to minimise use of the keyboard. Windows Vista shipped with reasonable voice recognition software. Windows 7 dramatically improved this software to the point where it is as good as most commercial voice recognition tools. This post will give my impression of using Windows Speech Recognition for the last month or so.

To activate speech recognition go to the control panel and choose "Ease of Access". The first thing you will have to do is go through a 20 minute speech training programme. This helps to software to recognise your voice. After that you are ready to go, or rather, ready to talk! I'm not going to go into details of how to use speech recognition as this is documented in the help files of the speech recognition software or on Microsoft's speech recognition website.

There are essentially two ways in which you use speech recognition software. You will either use it to dictate sentences to the computer, for example when writing an email. Or, you will use it to command your computer to take certain actions such as opening windows or browsing the web.


When you first start dictating sentences to the computer you will be frustrated. It is likely the software will only recognise about 50% of your words. This is simply not enough to be usable. However, do not to despair. The software learns as it goes along and quickly improves to the point where it does become useful. And eventually it will become quicker to dictate to your computer than you type your text manually. For me, this took about one month - and I am a very quick typist.

The speech recognition software has in built functions to help you correct what it types. It is very important that you use these functions and do not simply correct its mistakes with the keyboard. By correcting mistakes with the keyboard you will prevent the software from learning from them.

One interesting side effect of dictating to your computer is that it does make you think harder about what you are writing. Often typing emails is little more than a brain dump rather than meaningful text. It certainly helps me in writing more thoughtful text.

A word about how to dictate to your computer. There is a temptation when dictating to the computer to speak slowly and clearly and over pronounce words. This is counterproductive. It is best to speak in a normal voice than to change the way you speak. Obviously if you speak very fast you might need to slow down a little. The other thing, which I'm just getting used to now, is that you can actually dictate very long sentences in one go. Sometimes you will have to go back and correct words but this is generally quicker than dictating three words at a time.


Commanding the computer to do certain things also works OK, although sometimes it is frustrating because you are not sure how to command the computer using your voice. Certainly, knowing your keyboard shortcuts will help you get around your PC more easily, as you can command these by voice too. My previous entry is a good starting point for learning slightly less common keyboard shortcuts.

The other frustrating thing about commanding your PC is that generally speech recognition works better for sentences than it does for individual words. Thus I still sometimes find myself repeating the same command 15 times before finally giving up and grabbing for the mouse. However, even this is getting better as I start to get more familiar with how to command my PC.

Specific frustrations

Other than the more general issues mentioned above there are some specific frustrations that I have about the speech recognition software as it is.

Mouse grid

Some programs are simply too complicated to get around using voice commands. In such situations this speech recognition software provides something called "mousegrid". This lets you zoom in on the exact location that you want to click the mouse using voice commands. However Microsoft's implementation of this is rather clunky. Thankfully there is an excellent add on called Voice Finger. This provides a much more useful grid as well as other excellent features. I would certainly recommend using it if you're serious about using speech recognition.


Firefox and windows speech recognition don't mix very well. I think this is more the fault of Firefox than windows. The result is that Firefox as well as your computer as a whole become slow and almost unusable after a while. This is frustrating for me because I am a great Firefox fan. The problem appears to be the way in which Firefox presents the available links to the speech recognition software. Thankfully there is a way of disabling this for Firefox. This means you can still use voice recognition whilst having Firefox open, but commanding Firefox using voice recognition will be limited. Click here for more details about how to disable speech recognition in Firefox.

Background noise

Obviously background noise is more a function of your working environment and the quality of your microphone than of the speech recognition software. However there are things which are still frustrating. I work in a reasonably quiet environment but the voice recognition software often picks up noises such as typing on the keyboard or just simply knocking things and converts these to simple words such as "at" and "if". If I turn down the microphone's sensitivity eventually it has difficulty hearing my voice. So I haven't quite solved this problem yet but I do appreciate that this is probably more to do with the set up of my computer than with the speech recognition software.

Integrating with other software

By and large I have found that speech recognition integrates very well with other software, especially with Microsoft's own software. This is to be expected I guess, but still kudos to Microsoft for providing such good speech recognition software with windows 7.


Overall, windows and speech recognition is a high quality product that has helped me minimise the use of my keyboard. And just in case you're wondering, yes, I have dictated this whole post using windows speech recognition.