Speech Input and Output for Commercial Information Systems
Uploaded by Ethan on Nov 09, 2002
Abstract
Conventional user input devices such as keyboards and mice are not natural or necessarily intuitive and are often detrimental to the efficiency of a computer user. Complex user interfaces can slow a user’s progress and prolonged use of keyboards, mice and monitors can result in muscle strain and permanent injury. The learning curve of these traditional devices is also steep for new users – learning where the keys are, learning the right amount of power required to move a mouse, how fast to double-click and counter-intuitive software interfaces make it very difficult for a user to effectively use a computer without investing time in learning how to operate it.
Since the natural communication method for humans is speech, it makes sense that speech be investigated as a possible input and output method. This paper discusses the technology of speech recognition as an input method and speech synthesis as an output method; how the technology is currently being used in commercial systems and possible applications in the future.
Introduction
The primary reason speech is being researched as an input and output device is convenience. Current computers are controlled with keyboard and mice, limiting users to the desktop environment where the computer and processor physically exists (“Microsoft Research,” 2001). If the computer could be controlled without these physical interfaces it could be used anywhere and output information to the user at any time, in the user’s native language. Speech-controlled devices also make irrelevant user interface usability issues (McKay, 2001) and lessen the learning curve of a new computer user. Accessibility for disabled users will cease to be an issue with speech devices; no longer will fine coordination and vision be required to effectively operate a computer.
The aim of speech interface research is to allow computers to work alongside the user, making computing less time consuming and more productive. But for this to work effectively, speech recognition and synthesis needs to be combined with the artificial intelligence (and appropriate network infrastructure) to understand natural language, allowing users to simply ‘ask’ the computer to do something for them, rather than issuing a string of commands. The use of artificial intelligence to learn to understand what a particular user wants and means is what makes this approach to computing so effective and desirable.
Speech Recognition
Speech recognition, taking voice input and interpreting it either as commands or converting to text, has been the topic of research for many...