ABSTRACT

VIRTUAL WORLD COMMAND AND OUTPUT EXPERIMENT

by
Lynne Alison Winner

A 2 X 3 experiment with command interface (speech or keyboard) and output (text, voice, or both) was conducted. There were two analyses. The commanding analysis of time and preference was between subjects. The output analysis of accuracy and preference was within subjects. All subjects were briefly trained in a sample world and then asked to navigate within a three object virtual world. Objects were queried with either speech or keypress and output was given in text display, recorded voice, or both. Output type associated with each object was varied across subjects, canceling any preference or possible order effect. Sessions were timed to determine effectiveness of speech or keypress. A questionnaire was given to each subject after the experiment, assessing correct recall of object provided information. Subjective preferences on the experience were also sought. Experimental results yielded no significant effects. Some marginal effects were found. Higher accuracy of answers for subjects using keypress command was noted. Keypress was also the preferred command mode, although sessions with speech took a shorter time.

This report examines speech and virtual reality in the role of naturalistic interfaces. A recount of the experiment’s development and the factors involved in its outcome are also discussed.