Voice Profiles

Each user develops a voice profile that is a comprehensive collection of data regarding their vocal characteristics and language usage.  The profiles are first established during enrollment into the speech software.  Subsequently, if the proper techniques are performed the profiles adapt and produce better results.  With the software configured properly and best practices being used for dictation, 98% accuracy and above should be attainable.

Direct Dictation Model

Enrollment

• Initiated at one of the PC's where the software is loaded.
• Process consists of adding the user name and picking the closest available medical specialty language model.
• User reads sample text for approximately 7 minutes.  Although Nuance's Dragon Version 9 marketing material indicates that this is not required anymore, PCA strongly advises that the time be taken to complete this process.

Adaptation

• User profiles are highly adaptable for language, words, and commands.
• Adaptation occurs during regular usage following best practices. Training is crucial
• Adaptation can be forced through tools provided in the software.

Storage and Access

• Access can come from a local machine or server.
• Roaming capability refers to a voice profile stored on a network for the purpose of having it available to the end user on whatever PC loaded with the software they choose.
• Roaming requires placement on a server with the PC having a mapping assigned to the location.
• Roaming profiles get cached between the server and local client pc when opening and closing the program.  Approximately 15MB of data is pushed through the network during the caching process.  Subsequently, little network traffic is generated.
• Users can be disconnected from the network and still have access to their voice profiles.
• Backup is critical in case of voice profile corruption.  Automatic backups can be scheduled to a network location.