Another suggestion:

When creating a new user for the student, you might want to choose the "Teen"
voice model. I have used it with adults who have soft, high-pitched, or
somewhat quiet voices.

And a comment:

>her speech is good enough that Dragon should be able to understand what
>she's saying if it is trained enough.

In my experience, voice files in NaturallySpeaking Versions 4, 5, and 6 do not
become significantly more accurate with lots of training. Accuracy is
excellent after completing initial training and running the Vocabulary
Builder. Performance does improve with use; but this is due less to
improvements to the voice files than to improvements in the user's
understanding of how to dictate and operate software by voice. So I hope this
student is scheduled to receive many hours of training using
NaturallySpeaking, if she is able to get through enrolment!

Alan

Alan Cantor
Cantor + Associates Inc.
Workplace Accommodation Consultants
[log in to unmask]
www.interlog.com/~acantor