11.5

Chapter 1: Introduction to Dragon 11.5
Using multiple acoustic models with a
User Profile
If the computer that a user dictates on meets certain system requirements, Dragon 11 or greater
uses two acoustic models with a User Profile. Using two acoustic models instead of one improves
recognition and dictation accuracy. Acoustic models that can be added in pairs to a User Profile
are called BestMatch IV models.
Any time you add a new dictation source to a User Profile that uses two BestMatch IV models,
the dictation source will also be associated with the two models.
A User Profile with two BestMatch IV models uses more computer resources, including RAM
memory, than a User Profile with one acoustic model. Dragon 11 or greater uses two BestMatch
IV models on a multi-core computer with at least 2 GB RAM. Dragon 11 or greater recognizes
when it is installed on a computer that meets these system requirements and selects the Best-
Match IV acoustic model as the default.
Acoustic Optimization for User Profiles with
BestMatch IV acoustic models
If a User Profile includes BestMatch IV models, you must run the Acoustic and Language Model
Optimizer (ACO) on a computer with multi-core processors. Dragon 11 or greater uses two Best-
Match IV models only on a multi-core computer with at least 2 GB of RAM.
If the ACO processes a User Profile that was created on a computer with multi-core processors,
the optimizer always selects the BestMatch IV models to optimize.
If a User Profile is associated with BestMatch III models without accents, and you run the opti-
mizer on a computer that supports BestMatch IV models, if automatic acoustic model selection is
enabled, the optimizer will automatically choose a BestMatch IV model to optimize.
Multiple acoustic models and User Profiles on single
and multi-core computers
If a user creates a User Profile on a multi-core computer, when the user opens the User Profile on
a single-core computer, Dragon uses the first acoustic model for recognition. Dragon does not
load the second acoustic model.
If a user creates a User Profile on a single-core computer, when the user opens the User Profile on
a multi-core computer, Dragon uses the single acoustic model for recognition.
If a user selects BestMatch IV models for a User Profile and trains the profile on a multi-core com-
puter, when the user opens the User Profile on a single-core computer, Dragon displays the fol-
lowing warning message:
"Your computer has a single core processor. You have opened a User Profile created on a com-
puter with a multi-core processor.
15