How We Test an Application

The basic idea of usability testing is to gather together a small group of people (say, 10 to 15) who represent the population that will use the application. We then ask them to perform a series of tasks, derived from the objectives set out in the Requirements Specification. We observe them as they perform these tasks , noting where they have problems and success. We then correct the problems.

Usability testing is an art in itself ”and there are many good books available that explain the methods and processes of usability testing, ^[1] so we won't go into all the details here. However, most of these books focus on testing graphical user interfaces for computer applications, not audio user interfaces, such as those found in speech-recognition applications ”and there are some significant differences between them.

^[1] A great book is Usability Engineering by Jakob Nielsen (San Francisco: Morgan Kaufmann Publishers, 1994).

For example, when testing speech systems, we can't have test subjects speak out loud to give their feedback while they're using the system, because that could cause the speech recognizer to malfunction. Instead, we need to look for other reactions , such as body language, to determine when they're having problems.