Title :
A voice input word processor system
Author :
Leng, Goh Wee ; Mital, Dinesh P.
Author_Institution :
Sch. of Electr. & Electron. Eng., Nanyang Technol. Univ., Singapore
fDate :
11/1/1992 12:00:00 AM
Abstract :
An intelligent voice-based word recognition and processing system using natural language processing techniques is discussed. The system comprises three major components: the voice recognition module, the natural language processing module and the system integration module. A commercially available voice card is modified and interfaced to an IBM compatible personal computer (PC-AT), which also acts as a word processor. The system is capable of recognizing 6400 words. However, due to relatively long search times, a vocabulary of about 1000 words has been implemented. The natural language module comprises a parser and an understander submodule which performs critical lexical analysis to build bounded morphemes from its free forms with regard to the context of the sentence, and to generate the appropriate responses to reflect the consequence of the parse. The system is speaker dependent. Accuracies of up to 80% are achievable with an input string duration of up to 7 s
Keywords :
IBM compatible machines; add-on boards; dictation; microcomputer applications; natural language interfaces; speech recognition; word processing; IBM compatible personal computer; PC-AT; bounded morphemes; context; free forms; input string duration; lexical analysis; natural language processing techniques; parser; search times; speaker dependent system; system integration module; understander submodule; voice card; voice input word processor system; voice recognition module; Automatic speech recognition; Consumer electronics; Dictionaries; Handheld computers; Hardware; Microphones; Natural language processing; Natural languages; Speech recognition; Vocabulary;
Journal_Title :
Consumer Electronics, IEEE Transactions on