Voice Recognition Software

At various times over the past several years, I’ve tested the use of voice recognition software for dictating text, such as Dragon Naturally Speaking. Since I can speak much faster than I can type, the potential productivity benefits are too great to ignore. However, up until now I’ve always been disappointed by the results. The software I tried was always too inaccurate to be of much practical use even after a great deal of time spent training it to recognize my voice. I ended up spending more time correcting errors than I was saving. Regular typing was still faster.

Last week I began playing around with the built-in voice recognition capabilities of Windows Vista. It didn’t cost me anything extra, so I didn’t hold out much hope that it would be any good. I was pleasantly surprised. Training took only a few minutes, and the software’s accuracy was higher than anything I tried before. The interface and commands were intuitive and flexible. It was also nice and speedy on my laptop. I’ve been using the software to work on my book and to answer emails.

The software does a good job of learning from its mistakes. When it makes a mistake, you simply tell it what words to correct, and then restate the correct words. It then displays a list of choices that closely match what you say. Most of the time the correct choice is in the number-one slot. Once you’ve corrected a mistake, the software tends not to make that mistake again. You can also teach it new words and phrases.

Vista’s voice recognition certainly isn’t perfect, but I think it just crosses the line of productivity improvement. It still makes some silly errors, but it does a great job of choosing the correct word based on the context and doesn’t make too many mistakes with words like they’re and there or you’re and your.

The software can scan your existing emails and documents to familiarize itself with your vocabulary. This allowed it to correctly spell my name and other non-dictionary words the first time I said them. This helps the software behave fairly intelligently right out of the box. I assume that as Microsoft continues to make improvements to its voice recognition, the auto-update feature in Windows can install them.

The downside of using Vista’s built-in voice recognition capabilities is that it favors the use of Microsoft’s software. This isn’t a big deal for me, but it might be an issue for some. Outside of Microsoft software, the voice recognition features tend to be a little unreliable. For example, I can’t use it to dictate directly into The Journal. I can dictate into my blogging software, but the editing behavior is a little funky. It definitely works best when dictating directly into Microsoft programs. Usually I’ll dictate text directly into WordPad, and then I’ll copy and paste the text into another application. This is only a minor annoyance.

In addition to dictating text, you can also use voice recognition to navigate applications and even to browse the web. For the most part, I find this slower than using a mouse and keyboard, but it could be of great benefit to people with certain disabilities. It was kind of fun to navigate my website using only my voice though… at least in a very geeky sort of way. Of course when I tried to show it off to Erin, the voice recognition misunderstood every other word.

If you already use Windows Vista, I encourage you to check out the built-in voice recognition capabilities to see if they work well for you. To get started, simply click the Start button in Windows Vista, and then type “windows speech recognition” in the search box. Select the program of the same name, and go from there.

It’s nice to see that basic voice recognition is finally coming of age. This is an exciting time to be alive.

By the way this blog entry and the one before it were both made using Windows Vista’s speech recognition.