KR960025319A

KR960025319A - Automatic Learning Training Device in Speech Recognition System

Info

Publication number: KR960025319A
Application number: KR1019940038109A
Authority: KR
Inventors: 이용일
Original assignee: 박정옥; 주식회사 에스원
Priority date: 1994-12-28
Filing date: 1994-12-28
Publication date: 1996-07-20

Abstract

1. 청구범위에 기재된 발명이 속한 기술분야1. TECHNICAL FIELD OF THE INVENTION

음성인식시스템에 있어서 학습 훈련장치에 관한 것으로, 특히 자동으로 학습 훈련시켜 음성으로 저장한 후 재생하는 자동 학습 훈련장치에 관한 것이다.The present invention relates to a learning training apparatus in a speech recognition system, and more particularly, to an automatic learning training apparatus that automatically learns training and stores it as a voice and plays it back.

2. 발명이 해결하려고 하는 기술적 과제2. The technical problem to be solved by the invention

음성인식율을 높이고 음성등록 시간을 절약하는 시스템.A system that improves voice recognition rate and saves voice registration time.

3. 발명의 해결방법의 요지3. Summary of Solution to Invention

종래의 히든 마르코프모델(HIDDEN MARKOV MODEL)을 사용한 음성인식시스템에는 인식률을 높이기 위해서 매 단어 등록시 10회 이상의 학습훈련을 시키고, 시간이 많이 소요되는 등록과정을 한번의 정상적인 발성을 저장하여 자동으로 다양한 패턴을 만들어 재생시킴으로서 학습훈련에 필요한 음성 데이터를 만들기 위한 시스템을 제공함.The conventional voice recognition system using the Hidden Markov Model (HIDDEN MARKOV MODEL) has more than 10 training sessions for every word registration in order to increase the recognition rate, and saves a time-consuming registration process and saves one normal voice automatically. Providing a system for creating voice data required for learning and training by creating and playing back music.

4. 발명의 중요한 용도.4. Important uses of the invention.

학습 훈련장치.Learning Training Device.

Description

Automatic Learning Training Device in Speech Recognition System

본 내용은 요부공개 건이므로 전문내용을 수록하지 않았음Since this is an open matter, no full text was included.

제1도는 종래의 음성인식시스템도, 제2도는 본 발명에 따른 음성인식시스템도.1 is a conventional voice recognition system diagram, Figure 2 is a voice recognition system according to the present invention.

Claims

In the speech recognition system, a microphone 101 for converting a voice signal into a signal electrically, an amplifier 102 for amplifying the output signal of the microphone 101, and an A for digitizing the output signal of the amplifier 102. The amplifier 102 by analogizing the output of the / D converter 103, the speech processing circuit 119 for analyzing and synthesizing the output of the A / D converter 103, and the speech processing circuit 119; D / A converter 111 to provide back to), SRAM 110 to store the data to be generated and synthesized in the speech processing circuit 119, system controller 112 for controlling the whole for speech recognition and A digital signal for recognizing speech by extracting and learning and training a feature under control of the system controller 112 to generate a basic pattern, and comparing the output of the A / D converter 103 with a reference pattern of the input speech. Generated by the reasoning unit 114 and the digital signal processing unit 114. Auto-learn training device according to a speech recognition system as SRAM (113) and for storing a reference pattern value, characterized by consisting of a recognition result output unit 118 to output the speech recognition by the digital signal processor 114.

※ Note: The disclosure is based on the initial application.