10.0
Copyright © 2010 Nuance Communications, Inc. All rights reserved
19
Customizing the Vocabulary from Your E-Mail
As you learned, Dragon can adapt its Vocabulary based on designated documents. It also has the ability to
quickly learn from your usage of several e-mail programs: Lotus Notes, Microsoft Outlook, Outlook Express, and
Windows Mail. By “studying” the e-mails you sent as well as the recipient names, Dragon can identify
potentially useful contact names to add to the Vocabulary, as well as adapt its statistical information on word
usage (frequency and context).
You should take advantage of this tool as soon as possible after creating your User profile.
Step 1: Say open Accuracy Center, then say or click the link Increase accuracy from e-mail.
Step 2: The tool will detect supported e-mail programs installed on your computer and display their names
(this may take a moment). Check the box of the e-mail program(s) you want Dragon to analyze here.
Step 3: Use the checkboxes to indicate whether you want Dragon to only look for contact names to potentially
add to the Vocabulary, or whether you also want Dragon to adapt its statistics based on your sent mail.
The option to “Improve from my
e-mail writing style” is
recommended EXCEPT if what
you sent in that email program
contains another language.
Note: The analysis may keep
Dragon busy for a while,
especially the first time; later
on, it will go much faster if you
choose the option button
“Consider e-mail written since
my last scan.”
IMPORTANT: When you start this process, your e-mail program may bring up a message box for you to
allow Dragon to access it; if you have several windows open, this message may not be visible initially because it
is under another window.
Step 4: After its analysis, the tool will display what it identified as potentially useful additions.
The rest of the process is similar to Add Words from Documents: you can uncheck items, edit items,
and train items acoustically if you wish, then Dragon adds to the Vocabulary the checked items, plus
adapts its model to the word frequencies observed, if you opted for that.










