11.5

Acoustic and Language Model Optimization

Vocabulary customization and proper correction of recognition errors both help Dragon recognize your

dictations more accurately. “Accuracy Tuning” is a functionality you can use to further “tune” your profile’s

accuracy for your specific usage. It has two aspects:

 It can refine your User profile using the latest acoustic data archived from any corrections and acoustic

training you performed.

 It can use text from your previous dictations to add frequent word sequences to the Language Model

-- statistics Dragon uses to “predict” which words and spellings it should write when you dictate.

IMPORTANT: If you have recently dictated in a very different sound environment, such as an airplane, you

should not let Dragon perform acoustic optimization.

The Accuracy Tuning process becomes available once you have used Dragon for a while (the software builds up

an archive when you perform dictations, corrections and training). A message box will let you know once

Dragon has enough for its automatic optimization.

How much of an accuracy boost you will get from Accuracy Tuning will depend on factors such as the amount

and quality of the initial training and optimization (if performed during profile creation), the initial accuracy

you experienced, and the amount and quality of text and audio data available for processing.

You can choose to schedule the two components, Acoustic Optimization and Language Model Optimization, to

run periodically at a time and frequency of your choice. You can also launch one or both yourself, as shown

below.

Launching the Acoustic and/or Language Model Optimization

Step 1: Use the DragonBar’s Audio menu (or the Accuracy Center’s link “Launch Accuracy Tuning now”).