Dragon’s speech recognition Deep Learning process mean it’s constantly learning from you. By simply dictating, you’re giving it sound information and usage data that helps it get better at knowing how to transcribe effectively. Accuracy and dumping speed are critical when it comes to ensuring you’re as productive as possible. The software gets more accurate when you correct it, identifying misrecognitions and instructing the programme to essentially get better at guessing. It doesn’t understand English, so it can be quite tricky to decide which word to spit out when it hears homonyms or words with soundwave profiles that look very similar. Analysing a vocabulary of 100s of thousands of words, it’s about narrowing down which words form part of your core vocabulary and figuring out your syntax.
Another way Dragon improves its accuracy is by processing collected audio. Dragon’s acoustic optimisation process is part of your profile maintenance plan. It collects valuable sound information in Dragon’s own special format documents, which contain sound and text information. This is then verified by the optimiser and the resultant learning is streamlined into your voice profile for better accuracy. After a few weeks and months even, your voice profile encompasses all of this time spent customising and learning, which is what makes it quite valuable.
It’s easy enough to start a new voice profile and by today’s standards, you’re leaps and bounds ahead of previous editions of the software, which required much more customisation and tinkering to get near 99% results. However, it’s the little things that make the difference and saves some frustration in covering old niggles. Dragon is incredibly accurate out-of-the-box, usable from the first minute but the real time-saving happens when you don’t have to correct as much.
Be careful. Your accuracy will get to a point where you’re so confident in Dragon’s ability to interpret your voice that your proof-checking will become less important. While Dragon’s an amazing tool, it’s not infallible and you may have the odd misrecognition creep into your work.