The Art and Business of Speech Recognition: Creating the Noble Voice
Designers are often tempted to make every menu a list of options or commands, particularly when there are four to six items from which to choose. However, if callers are in part of an application they may use only once in a while, we don't want them to have to think about remembering a lot of commands. Often applications have a preferences section, in which users can change some settings of how the application works. In a Web browser the preferences section would allow the user to change the default home page. In a speech system the preferences section might allow users to change things like their PIN numbers , how the system provides certain types of information, and so on. Since users seldom need to access the preferences section of an application, they won't have the opportunity to learn the commands of the possible choices. A bad example of how a prompt for a preferences section would be "Would you like to change 'How I read stock information,' 'How verbose the prompts are,' 'Your PIN number,' or 'Your personal information?'" A better alternative is to walk callers through a process that is slightly longer but does not require them to learn a single command. It also saves designers from having to write extremely compact commands. Here's how it works.
Reducing everything to a "yes or no" question makes the interaction much easier for both the designer and the caller. This method is particularly useful in parts of the application that are less frequently used and when the options in question don't form a cohesive group . |