The SpeechWear project examines how speech might be used as a basis for interaction between a user and a small portable computer. As computer become smaller and more ubiquitous, and as people become accustomed to accessing computing resources in many different environments, traditional interface modalities (based on keyboards and monitors) will have to be replaced with modalities that do not depend on such equipment. Speech interaction is a natural choice for this role.
Speech interaction, however, will have to rapidly move beyond its current paradigms (based on existing interfaces) and become much more like human-human interaction in the ways that information is transmitted. Computer system will need to possess specific conversational skills similar to those learned by those humans who have successfully mastered the conventions governing the way that we talk to one another.
As an initial demonstration of a speech-based wearable device, we have implemented a system that allows Marine mechanics to complete a vehicle inspection using speech. The system provides a speech interface to a hypertext browser (Mosaic) and allows users to navigate and enter data, for a task that previously used a pencil and paper format. The interface provides access to both locally stored information (e.g., the inspection form) and though a wireless link, to information (e..g, documentation or repair procedures) available on remote servers.
SpeechWear was field tested at Camp Pendleton and is currently undergoing further evolution, in preparation for deployment as part of a maintenance information system.
A paper describing this work is available, in either html or postscript format.
More information may be obtained from Alexander Rudnicky.