An Integrated Multimedia Environment for Speech Recognition Using Handwriting and Written G.pdf

Proceedings of the 36th Hawaii International Conference on System Sciences - 2003An Integrated Multimedia Environment for Speech Recognition Using Handwriting and Written Gestures Wu-chun Feng feng@ Computer Computational Sciences Division Los Alamos National Laboratory P.O. Box 1663 Los Alamos, NM 87545 Abstract To provide a natural interface to the computer, we present an integrated speech, gesture, and handwriting recognition system. By integrating these technologies, we can easily accomplish all the tasks that are needed to use computers and their applications, including many tasks which neither technology can adequately perform by itself. This integrated environment not only makes it easier for novices to use computers but also increases the productivity of experienced computer users. 1. Introduction Computers have quickly become a ubiquitous part of society. While many of us in the engineering and science disciplines take our computer skills for granted, a large number of people still have little to no experience using a computer. To complicate matters, those people who want to learn how to use a computer often do not have the time or patience to become skilled with standard types of computer input devices such as the keyboard, mouse, or trackball. To ease the transition, speech can be used as a natural way to interface with the computer. A speech interface can also benefit users who are familiar with standard types of input devices. Since nearly everyone can talk faster than they can type, speech can increase productivity if (1) the speech recognizer is fast and nearly error-free, e.g., 95% recognition accuracy, and (2) the tools for correcting speech- recognition errors are simple and natural to use, are seamlessly integrated with the speech recognizer, and are nearly error-free, e.g., 99% recognition accuracy). The high-recognition rate for the speech recognizer and even higher-recognition