SOUND INFORMATION DESCRIBING METHOD AND SOUND INFORMATION RECOGNIZING METHOD

PROBLEM TO BE SOLVED: To easily describe sound information in a form excellent in a general applicability and to recognize voices using a conventional voice recognition device in a sound information describing method and a sound information recognition method. SOLUTION: First, the sound information...

Full description

Saved in:
Bibliographic Details
Main Authors SAGAYAMA SHIGEKI, SANEHIRO TAKATOSHI
Format Patent
LanguageEnglish
Published 04.04.1997
Edition6
Subjects
ACOUSTICS
MUSICAL INSTRUMENTS
PHYSICS
SPEECH ANALYSIS OR SYNTHESIS
SPEECH OR AUDIO CODING OR DECODING
SPEECH OR VOICE PROCESSING
SPEECH RECOGNITION
Online AccessGet full text

Cover

More Information
Summary:PROBLEM TO BE SOLVED: To easily describe sound information in a form excellent in a general applicability and to recognize voices using a conventional voice recognition device in a sound information describing method and a sound information recognition method. SOLUTION: First, the sound information to be converted is recognized based on a voice model 2 that is beforehand generated employing a vocabulary no- limitation voice recognition device 1 which generates an arbitrary sound syllable column or a recognition unit column such as a phoneme column as a recognition candidate. Then, the obtained recognition result is registered into a word list 4 which is used for the recognition in a vocabulary limited voice recognition device 3. The description of the sound information obtained in the above process represents the recognition word column that is closest (similar) to the sound information in the model 2 being posessed by the device 3. Therefore, the collation between an inputted voice and the sound information is conducted by using the device 3 based on the list 4 generated by the above procedures.
Bibliography:Application Number: JP19950249864