National Changhua University of Education Institutional Repository : Item 987654321/18399

English | 正體中文 | 简体中文 | Items with full text/Total items : 6507/11669
Visitors : 29956872 Online Users : 530

RC Version 3.2 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.

Scope Adv. Search

Login ‧ Upload ‧ Help ‧ About ‧ Administer

NCUEIR > College of Technology > vr > NSC Projects > Item 987654321/18399

Please use this identifier to cite or link to this item: http://ir.ncue.edu.tw/ir/handle/987654321/18399

Title:	應用語音訊號之經驗模態分解與魏格納分佈技巧之語者辨識系統
Authors:	吳建達
Contributors:	車輛科技研究所
Keywords:	語者辨識系統;經驗模態分解;Wigner-Ville 分解;能量頻譜;倒傳遞類神經;廣義回歸類神經 Speaker identification;Empirical mode decomposition;Wigner-Ville distribution;Energy spectrum;Backpropagation neural network;Generalized regression neural network
Date:	2011
Issue Date:	2014-04-29T07:29:21Z
Publisher:	行政院國家科學委員會
Abstract:	本研究計畫主要是發展以語音訊號為基礎，並搭配經驗模態與Wigner-Ville 分解的方法並結合類神經系統所建構成的語者辨識系統。研究工作包括: 語者聲音訊號的錄製、語音訊號的探討、經驗模態與Wigner-Ville 分解理論之實現、智慧型分類技巧的適用評估比較。在實驗方面，首先利用麥克風、擷取資料系統進行語音錄製，邀請男女生36 人進行錄音動作，來做為語者辨識之資料庫。在特徵擷取方面，分別以經驗模態分解和Wigner-Ville 分解技術作為特徵擷取之基礎，經驗模態分解在訊號篩選的過程中不會失去原始訊號的特性而且可獲得有效本質的模態函數元件，以計算各元件的能量；之後分別依照不同的能量分布當特徵，並結合倒傳遞類神經與廣義回歸類神經網路作分類，進而達到語者辨識的目的。 In this project, an intelligent system of speaker identification and classification based on speech signal with Empirical Mode Decomposition (EMD), Wigner-Ville distribution (WVD) and neural networks system is studied. These speech signals are measured using microphone and data acquisition system from some individual speakers speaking five Chinese short sentences. The feature extraction based on EMD and WVD with energy spectrum will be used. The EMD sifts the complex signal of time series without losing its original properties and then obtains some useful intrinsic mode function (IMF) components. Calculating energy of each component can reduce the computation dimensions and enhance the performance of classification. The WVD has received considerable attention in recent years as an analysis tool for non-stationary or time-varying signals. The WVD furnishes a high resolution and instantaneous energy density spectrum in the time and frequency domains. In the speaker identification, the back-propagation neural network (BPNN) and generalized regression neural network (GRNN) will be applied to verify the performances and the training time in the proposed system.
Relation:	國科會計畫, 計畫編號: NSC100-2221-E018-009; 研究期間: 100/08-101/07
Appears in Collections:	[vr] NSC Projects

Files in This Item:

File	Size	Format
index.html	0Kb	HTML	391	View/Open

Loading...