Please use this identifier to cite or link to this item:
標題: 環境控制輔具之語音辨識系統設計
Speech Recognition System Design of Environment Control Auxiliary Equipment
作者: 王煌仁
Wang, Huang-Jen
關鍵字: Speech Recognition
Hidden Markov Model
Microhpone Array
出版社: 電機工程學系
摘要: 中文摘要 本論文為國科會環境控制輔具計畫之語音控制的主要部份,其目的是發展一套適合行動不便之病患,以語音辨識的方式來控制其生活周遭的家電用品,以營造舒適的生活機能。 在本文中,我們設計了兩套系統,一套是使用一般麥克風的語音辨識系統,另外一套則為麥克風陣列語音辨識系統。 一般麥克風的語音辨識系統中,使用越零率、能量方法將有效語音段切割出來,而麥克風陣列語音辨識系統中,則計算出語音到達每個麥克風的延遲時間後,使用Delay and Sum Beamformer將語音訊號加強,最後兩個系統皆用MFCC擷取出語音的特徵參數,並使用隱藏式馬可夫模型做為語音辨識。
Abstract This paper develops a system for patients with physically disabling. With the use of speech recognition, the patient can control the electrical appliances conveniently in daily life. In this thesis, two systems of the speech recognition were designed. One isthe general microphone speech recognition system, and the other is the microphone array system. In contrast with the general microphone speech recognition system, the microphone array system calculates the delay time of each microphone and then use delay time and beamformer to enhance the speech signals rather than use the zero-crossing rate and energy to segment the speech signal. After that, the MFCC is used to extract the feature parameters of speech signals in both systems. Then the Hidden Markov Model is adopted to recognize the speech signals.
Appears in Collections:電機工程學系所



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.