This article reviews human factors research on the design of systems that use speech recognition for human control of the system or that use speech generation for the display of information. Speech technology terms are defined and the current status of the field is reviewed. Included are the performance of current speech recognition and generation algorithms, descriptions of several applications of the technology to particular tasks, and a discussion of research on design principles for speech interfaces. Finally, directions for further research are suggested. The need for better simulation techniques and performance measures is stressed, as is the importance of considering the entire system in which speech technology will function.

This paper captures high-leveldesign imperatives and solutions used to expand a two-sided model to an n-sided model to premiere interactive, multi-sided, coalitionwarfare simulations.We focus on general considerations with impact beyond the bounds of a spwific model.The designs discussed are inco~orated in version 1.85 of the Joint Theater Level Simulation (JTLS) and are being tested with a ten-sided Southwest Asia scenario.This U.S. Department of Defense model now supports a database-defined number of sides ranging from two to ten.Each side maintains an independent intelligence perception of the theater of operations and a designation of its friends, its enemies, and any neutrals. Multiple sides can align in "grand coalitions" or segregate into flexible groups to engage in multi-sided conflict.All sides are split into a variable number of dynamic factions. Each faction has specific attributes profiled by its Battlefield Operating Systems. Each faction holds a unique set of units and targets. And, they can subdivide, merge, or transfer to another side.

The purpose of this study was to investigate the extent t o which off-the-shelf voice recognition equipment can perform a s a speaker-independent system. independence" is meant the a b i l i t y of the system to be used by a larger number of individuals than t h a t which trained the system ( b u t including those who trained i t ) , and by individuals different from those who trained i t .

No abstract