國立高雄大學圖資館 |

Language: English

Back

結合基因演算法及經驗模態分解進行強健性語音辨識與FPGA晶片實現 = T...

國立高雄大學資訊工程學系碩士班

結合基因演算法及經驗模態分解進行強健性語音辨識與FPGA晶片實現 = The FPGA Implementation of Robust Speech Recognition System by Combining Genetic Algorithm and Empirical Mode Decomposition

Record Type:	Language materials, printed : monographic
Paralel Title:	The FPGA Implementation of Robust Speech Recognition System by Combining Genetic Algorithm and Empirical Mode Decomposition
Author:	李胥瑜,
Secondary Intellectual Responsibility:	國立高雄大學
Place of Publication:	[高雄市]
Published:	撰者;
Year of Publication:	2009[民98]
Description:	69面圖，表 : 30公分;
Subject:	FPGA
Subject:	EMD
Online resource:	http://handle.ncl.edu.tw/11296/ndltd/88636312804153602110
Summary:	本論文以實現一個強健性的語音辨識系統於FPGA平台上為主要目標。為了加快語音辨識處理的速度，以適合於速度較慢的嵌入式平台上，採用了整數快速傅立葉轉換取代浮點數快速傅立葉轉換，在辨識率些微下降下，大幅提高了快速傅立葉轉換的運算速度。對於含雜訊的語音，採用了黃鍔院士所提出的經驗模態分解法(Empirical Mode Decomposition, EMD)將含雜訊的語音訊號分離成多組本質模態函式(Intrinsic Mode Function, IMF)，針對不同本質模態函式做語音辨識上之實驗，以期能找出本質模態函式與語音雜訊間之關係，並以實數型基因演算法找出本質模態函式之權重，將分離出之本質模態函式依權重還原成語音後，能降低雜訊之干擾，提升語音辨識系統抗雜訊之能力。 In this thesis, we propose a robust speech recognition system constructed by utilizing HMM; the system utilizes Integer FFT to improve the computing performance and had been implemented in the FPGA embedded system. In the embedded system, the computing speed is not as fast as personal computer which would cause the speech recognition to consume more computing and power; in further, the requirement of real-time recognition would not be met. In the proposed method, we utilize Integer FFT to replace Float FFT and the experimental result shows that much computing time could be reduced while only a little recognition accuracy is lost. In addition, we also utilize Empirical Mode Decomposition to divide speech signals into multiple Intrinsic Mode Functions, and analyze the correlation between speech signal and noise in each Intrinsic Mode Function. Finally, we utilize Genetic Algorithm to find the optimal composing weight, and then, use it to reassemble the multiple Intrinsic Mode Functions to the speech signal for speech recognition.

結合基因演算法及經驗模態分解進行強健性語音辨識與FPGA晶片實現 = The FPGA Implementation of Robust Speech Recognition System by Combining Genetic Algorithm and Empirical Mode Decomposition
李, 胥瑜

結合基因演算法及經驗模態分解進行強健性語音辨識與FPGA晶片實現 = The FPGA Implementation of Robust Speech Recognition System by Combining Genetic Algorithm and Empirical Mode Decomposition / 李胥瑜撰 - [高雄市] : 撰者, 2009[民98]. - 69面 ; 圖，表 ; 30公分.
參考書目：面.
FPGAEMD

結合基因演算法及經驗模態分解進行強健性語音辨識與FPGA晶片實現 = The FPGA Implementation of Robust Speech Recognition System by Combining Genetic Algorithm and Empirical Mode Decomposition
LDR:03191nam a2200277 4500 001 244177
005 20170214095810.0
009 244177
010 0 $b 精裝
010 0 $b 平裝
100 $a 20170214d2009 m y0chiy09 ea
101 1 $a chi $d chi $d eng
102 $a tw
105 $a ak am 000yy
200 1 $a 結合基因演算法及經驗模態分解進行強健性語音辨識與FPGA晶片實現 $d The FPGA Implementation of Robust Speech Recognition System by Combining Genetic Algorithm and Empirical Mode Decomposition $f 李胥瑜撰
210 $a [高雄市] $c 撰者 $d 2009[民98]
215 0 $a 69面 $c 圖，表 $d 30公分
314 $a 指導教授：潘欣泰
320 $a 參考書目：面
328 $a 碩士論文--國立高雄大學資訊工程學系碩士班
330 $a 本論文以實現一個強健性的語音辨識系統於FPGA平台上為主要目標。為了加快語音辨識處理的速度，以適合於速度較慢的嵌入式平台上，採用了整數快速傅立葉轉換取代浮點數快速傅立葉轉換，在辨識率些微下降下，大幅提高了快速傅立葉轉換的運算速度。對於含雜訊的語音，採用了黃鍔院士所提出的經驗模態分解法(Empirical Mode Decomposition, EMD)將含雜訊的語音訊號分離成多組本質模態函式(Intrinsic Mode Function, IMF)，針對不同本質模態函式做語音辨識上之實驗，以期能找出本質模態函式與語音雜訊間之關係，並以實數型基因演算法找出本質模態函式之權重，將分離出之本質模態函式依權重還原成語音後，能降低雜訊之干擾，提升語音辨識系統抗雜訊之能力。 In this thesis, we propose a robust speech recognition system constructed by utilizing HMM; the system utilizes Integer FFT to improve the computing performance and had been implemented in the FPGA embedded system. In the embedded system, the computing speed is not as fast as personal computer which would cause the speech recognition to consume more computing and power; in further, the requirement of real-time recognition would not be met. In the proposed method, we utilize Integer FFT to replace Float FFT and the experimental result shows that much computing time could be reduced while only a little recognition accuracy is lost. In addition, we also utilize Empirical Mode Decomposition to divide speech signals into multiple Intrinsic Mode Functions, and analyze the correlation between speech signal and noise in each Intrinsic Mode Function. Finally, we utilize Genetic Algorithm to find the optimal composing weight, and then, use it to reassemble the multiple Intrinsic Mode Functions to the speech signal for speech recognition.
510 1 $a The FPGA Implementation of Robust Speech Recognition System by Combining Genetic Algorithm and Empirical Mode Decomposition
610 0 $a FPGA $a 基因演算法 $a 整數傅立葉轉換 $a 經驗模態分解 $a 語音辨識 $a 隱藏式馬可夫模型
610 1 $a EMD $a FPGA $a GA $a HMM $a integer fast Fourier transform $a speech recognition
681 $a 008M/0019 $b 464103 $v 2007年版
700 1 $a 李 $b 胥瑜 $4 撰 $3 387823
712 0 2 $a 國立高雄大學 $b 資訊工程學系碩士班 $3 353878
801 0 $a tw $b 國立高雄大學 $c 20100428 $g CCR
856 7 $2 http $u http://handle.ncl.edu.tw/11296/ndltd/88636312804153602110