May/June 2004

By Dr. James A. Larson

A user profile contains information that describes how to personalize a speech user interface to meet the needs for a specific user. For example, a user profile might contain the following information:

• The user’s preferred language. There are large segments of the population who cannot or prefer not to speak the local language. For example, about 15 percent of the population in the United Stated speaks Spanish, with 5 percent speaking only Spanish.

• An alternative acoustic model to accommodate a non-local speaker’s accent. Speech recognition engines often have difficulty understanding spoken words and phrases uttered by users with a strong accent.

• An increase in the default timeout limits, giving a user extra time to respond to prompts. This may be especially important for non-native speakers or users with impaired cognitive processes, such as patients with Alzheimer’s disease.

• Whether or not the user utilizes special text-entry devices, such as TTYs that enable users to enter text rather than speak.

• A notice that alerts the system if the caller uses a special device that enables them to enter text rather than speak.

• The preferred voice used by the speech synthesis engine.

• The user’s level of expertise – novice or expert.

• Protocols for identifying the user. Occasionally users will forget their passwords and personal identification numbers. These users must answer a series of questions for which only the user is likely to know the answer. Examples include what your mother’s maiden name is, or the year of your birth minus 18.

Many other parameter values may be stored in the user profile. An application should adjust its presentation based on data in the user profile. This extends the number of users who can effectively use the application.

Another way to extend the number of users who can successfully use an application is to create a standard key sequence to change the parameters in a user profile. For example, North American users press standard key sequences for special purposes: 911 for emergencies, 411 for directory service and 0 for operator. Another standard key sequence should be reserved for changing the parameters in the user profile. In the January/February 2003 issue of Speech Technology Magazine, the “Technology Trends” article summarized a basic set of universal commands for telephone users.

(http://www.speechtechmag.com/issues/8_1/cover/1583-1.html). One of these commands is “profile” (as well as the German, Spanish, Italian and French equivalent word) for changing pre-set parameters. Any user should be able to say “profile” to invoke a standard script for changing parameters in the user profile. Users should be able to invoke this standard dialog at any time, from any speech application, whenever they need to create or change user profile parameter values.

User profiles have proved popular and useful for many types of user interfaces. Speech user interfaces are no exception. Profiles allow users to personalize their user interfaces, which, in turn make users more comfortable and productive.

Dr. James Larson is Manager of Advanced Human Input/Output at Intel Corporation and author of the book, Voice XML – Introduction to Developing Speech Applications. He can be reached at jim@larson-tech.com and his Web site is http://www.larson-tech.com/.