VOICE RECOGNIZING SYSTEM AND PROGRAM THEREOF

PROBLEM TO BE SOLVED: To prevent voice data from being converted into wrong meaning in voice recognition by segmenting the voice data at a voiceless point per unit of recognizing the voice data and associating the recognized voice with text and storing them. SOLUTION: A recognition preprocessing uni...

Full description

Saved in:
Bibliographic Details
Main Authors MURAMATSU EIJI, TERADA MATSUAKI, ONO HIROKI, OSHIMA KOTA
Format Patent
LanguageEnglish
Published 18.08.2005
Edition7
Subjects
ACOUSTICS
ELECTRIC COMMUNICATION TECHNIQUE
ELECTRICITY
MUSICAL INSTRUMENTS
PHYSICS
SPEECH ANALYSIS OR SYNTHESIS
SPEECH OR AUDIO CODING OR DECODING
SPEECH OR VOICE PROCESSING
SPEECH RECOGNITION
TELEPHONIC COMMUNICATION
Online AccessGet full text

Cover

More Information
Summary:PROBLEM TO BE SOLVED: To prevent voice data from being converted into wrong meaning in voice recognition by segmenting the voice data at a voiceless point per unit of recognizing the voice data and associating the recognized voice with text and storing them. SOLUTION: A recognition preprocessing unit 71 segments the extracted voice stream packet at a voiceless section (α) to insert the voiceless data into a voiceless section (β) contained between the mute sections. Thus, a voice recognizing unit 72 performs voice recognition of the voice stream packet using the voiceless section (α) as a unit to convert the voice into text data, and stores the text data so that the data are associated with the voice stream packet before conversion via a recognition voice/text management section 73. The unit 71 also refers to the header information of the extracted voice stream packet to determine a portion having an interval of predetermined time or more as a segment from a time stamp included in the header information, and supplies the portion as a unit of voice recognition to the unit 72. COPYRIGHT: (C)2005,JPO&NCIPI
Bibliography:Application Number: JP20040029346