ENVIRONMENTAL ADAPTIVE MULTIPLE VOICE RECOGNITION APPARATUS AND VOICE RECOGNITION METHOD

摘要

PURPOSE: An environmental adaptive multiple voice recognition apparatus and a voice recognition method are provided to accurately judge a voice recognition result by extracting the optimum voice recognition result. CONSTITUTION: A voice input module(300) is formed of a microphone(310) receiving voice signals, and an analog-to-digital converter(320) converting the inputted analog voice signals into digital voice signals. A preprocessor(110) of a voice recognition module(100) analyses characteristics of the inputted digital signals for extracting characteristic vectors. The voice recognition module includes a plurality of recognition units(120) comparing the extracted characteristic vectors with a model which is previously stored in a memory by training for outputting voice recognition results. A judgement module(200) decides the optimum voice recognition result out of the voice recognition results.