Formerly known as Dragon NaturallySpeaking and now known just as Dragon, the speech recognition software has been around for over two decades. Each subsequent release of the software promises up to 99% accuracy, a figure which has become easier and easier to achieve over the years as the development team have refined algorithms and processing power has improved. While the vocabulary dwarfs any human’s in terms of word count, the software isn’t at the same point as Jarvis in Iron Man or the OS in Her.
Designed to transcribe speech, Dragon isn’t designed to understand speech. This may sound like an obvious statement, but it will give you some insights into how it operates. Moreover, you’ll be able to get better results when you consider the fact that it doesn’t understand the meaning of the words it’s dropping onto screen. As much as it could be the best listener in your workplace or home, simply converting your talk into type doesn’t mean it’s processing this the way humans could repeat back to you.
This means that speaking louder won’t help Dragon suddenly understand you. The same goes for speaking slower – something we often think will assist when there are communication barriers whilst talking to someone who speaks a different language. Think of Dragon interpreting soundwaves rather than meaning and it will help you understand why it struggles with smaller words and homonyms over longer words or phrases. Giving the software more sound information helps it distinguish between one word and another. That’s why its advisable to speak in phrases rather than one word at a time.
The best thing you can do when it comes to getting the best results from Dragon is to adopt a consistent pattern to your dictation. Siphon out inflection if you’re prone to dictating with sharp tonal shifts. Don’t be a robot but speak at a steady pace in a normal tone of voice. Copy the way newscasters read the news and Dragon will find it easier to match the words with the sounds.