| Abstract
|
Our objective was to build a speech interf … Our objective was to build a speech interface for use with a medicaldecision-support application. We used a commercially available speech-recognitionsystem to obtain textual representations of input utterances and we generatedgrammars to supply the anticipated utterances for the recognition system. Wedesigned the interface to identify controlledvocabulary terms from grammatical and ungrammatical textual transcriptions ofutterances. Our method included two steps: translation of the controlledvocabulary terms and textual representations of input utterances into keyword-based canonical forms, and matching canonical forms of input utterances tocanonical forms of the controlled vocabulary terms. Evaluation of the speechinterface indicated that this matching approach increased the identificationof terms from misrecognized utterances. ionof terms from misrecognized utterances.
|
| Author
|
Smadar Shiffman +,
William M. Detmer +,
Christopher D. Lane +,
Lawrence M. Fagan +
|
| Bibtype
|
techreport +
|
| Institution
|
Knowledge Systems, AI Laboratory +
|
| Key
|
KSL-94-45 +
|
| Modification dateThis property is a special property in this wiki.
|
1 May 2009 14:05:19 +
|
| Note
|
Updated February 1995. +
|
| Number
|
KSL-94-45 +
|
| Tag
|
Computer science +
|
| Title
|
A Continuous-Speech Interface to a Decision-Support System: I. Techniques to Accommodate for Misrecognized Input +
|
| Tr id
|
KSL-94-45 +
|
| Year
|
1994 +
|
| Categories |
Technical Report,
Publication,
KSL Technical Report
|
