Designing Effective Speech Interfaces

Command-and-Control

Command-and-control speech applications refer to a type of speech technology that recognizes a limited vocabulary of individual words and phrases spoken by a user. There are some special guidelines to follow when designing an interface for command-and-control applications.

Don't Use “I”

In a command-and-control interface, the system recognizes only certain commands and terms spoken by the human. If the computer uses the first person, “I,” users tend to talk in longer phrases, make more vague requests, and are more polite. These tendencies lead to more errors in systems that require specific commands.

Use Modeling

In a GUI interface, users are constrained as to their behavior—they can press only certain buttons or controls. In a speech interface, however, users can say anything (although the system will not always respond well). To constrain user speech to a form that is better understood and minimizes errors, use modeling to convey to the user the terminology and structure they should use. For example, this first script does not model constrained speech and results in unconstrained speech by the user:

System: “Welcome to ABC Travel Automated System. We look forward to servicing your travel needs. What are the dates of travel that you would like me to check for?”

User: “We are interested in traveling the first week of July, say July 1st to July 5th.”

If the speech engine can handle this type of continuous speech, then you will not have a problem. But in a command-and-control ...