| Paper: | 03-P1.9 (ICASSP 2003 Paper) | ||
| Session: | ICASSP 2003 Papers | ||
| Time: | Tuesday, May 18, 13:00 - 15:00 | ||
| Presentation: | Poster (ICASSP 2003 Presentation) | ||
| Topic: | ICASSP 2003 Papers: ICASSP 2003 Papers | ||
| Title: | ON THE USAGE OF AUTOMATIC VOICE RECOGNITION IN AN INTERACTIVE WEB BASED MEDICAL APPLICATION | ||
| Authors: | Claudio Eccher; ITC-irst | ||
| Lorenzo Eccher; ITC-irst | |||
| Daniele Falavigna; ITC-irst | |||
| Luca Nardelli; ITC-irst | |||
| Marco Orlandi; ITC-irst | |||
| Andrea Sboner; ITC-irst | |||
| Abstract: | The paper describes a multi-modal browsing system, developed by us,that allows to add automatic voice recognition functions to standardInternet browsers (e.g. ``Internet Explorer'' or ``Netscape''). Thebasic idea, underlying the system design, consists in the definitionand consequent realization of a software architecture capable ofhandling multi-modal interactions through the synchronization of HTMLand VoiceXML documents. HTML documents define a usual interactionbased on standard devices, such as: graphic monitor, keyboard, mouseand touch screen, while VoiceXML documents define a correspondinginteraction based on voice. The multi-modal browsing system detectsevents coming from various types of input devices, including amicrophone, and provides output according to predefined spatial andtemporal layouts. In general, the spatial layout is specified in HTMLdocuments, while the temporal layout is specified in VoiceXMLdocuments. In this way, the user can freely interact by using thepreferred device (e.g. the mouse, for selecting an item from a shortlist of options, or voice for filling the fields of a form) and thesystem is able to provide output in a coherent way.For both testing the system and tuning its parameters we have chosen amedical application scenario, where the goal is that of entering dataof laboratory test results into a patient database. To cope with thisapplication, namely a distributed Electronic Patient Record (EPR), wedeveloped a system in the past that utilizes a HTML based webinterface for accessing the patient database. Hence, to add voicebrowsing capabilities to the system, we only need to define a VoiceXMLinterface corresponding to the given HTML one (i.e. we must write acorresponding set of VoiceXML documents).In general, the usage of EPRs implies a large amount of work by peopleinside hospitals, for storing clinical data collected in differentformats: text reports, numerical values, images, etc. Moreover, EPRsallow the user to easily go across a huge amount of information tofind the needed one. Hence, EPRs are usually organized in sections,in which information is structured in categories that are homogeneousfrom the medical knowledge perspective. Whatever the EPR structureis, data entry and retrieval is primarily accomplished by usingkeyboard and mo | ||
| Back | |||
Home -||-
Organizing Committee -||-
Technical Committee -||-
Technical Program -||-
Plenaries
Paper Submission -||-
Special Sessions -||-
ITT -||-
Paper Review -||-
Exhibits -||-
Tutorials
Information -||-
Registration -||-
Travel Insurance -||-
Housing -||-
Workshops