(3.238.174.50) 您好!臺灣時間:2021/04/20 23:07
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果

詳目顯示:::

我願授權國圖
: 
twitterline
研究生:馮庭煜
研究生(外文):Ting-Yu Feng
論文名稱:以GM(1,1)模型為基礎的灰色語音活動檢測方法
論文名稱(外文):Grey Voice Activity Detection Based on GM(1,1) Model
指導教授:謝政勳謝政勳引用關係
指導教授(外文):Cheng-Hsiung Hsieh
學位類別:碩士
校院名稱:朝陽科技大學
系所名稱:資訊工程系碩士班
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2007
畢業學年度:95
語文別:中文
論文頁數:77
中文關鍵詞:G.729灰色頻譜削減法語音活動檢測GSM AMRGM(1.1)模型
外文關鍵詞:voice activity detectiongrey magnitude spectral subtractionG.729GSM AMRGM(1.1) model
相關次數:
  • 被引用被引用:0
  • 點閱點閱:380
  • 評分評分:系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔
  • 下載下載:26
  • 收藏至我的研究室書目清單書目收藏:0
一般語音的溝通中,經統計大約有28%是屬於非語音的狀態,也就是沒有語音信號的存在,此時若能利用技術正確地找出語音與非語音的部份,只需對語音的部份做進一步處理及使用簡單的雜訊編碼技術來模擬背景雜訊,就能有效節省頻寬的使用率,降低編碼的計算複雜度。語音活動檢測(voice activity detection, VAD)方法就是用來在摻雜的環境中正確判別出語音與非語音的一項重要前置技術。
目前,許多的語音活動檢測方法都假設輸入語音為外加信號模型(additive signal model),也就是說,輸入語音是由乾淨語音與外加雜訊合成的。因此,語音活動檢測的性能十分依賴外加雜訊的估計。適當的雜訊估計往往會導致好的結果。一般而言,語音活動檢測方法會根據估計的雜訊統計值建構語音活動檢測的決定法則(decision rule),然後利用決定法則判斷語音與非語音。在語音活動檢測的決定法則中,有一種是信號雜訊比(signal-to-noise ratio, SNR)。在外加信號模型的假設下,本論文利用GM(1,1)模型適當地估算出語音信號與外加雜訊的統計值,並以此資訊求出每一音段的SNR。最後透過適應調整門檻值(adaptive threshold)判斷輸入音段為語音或非語音。此方法我們稱為灰色語音檢測方法(grey voice activity detection, GVAD)
另一方面,在低信號雜訊比的的情況下,由於背景雜訊變化大,易造成語音活動檢測發生錯誤的判斷,本論文針對此問題亦提出一個以灰色頻譜削減法為基礎之能量語音檢測方法(grey magnitude spectral subtraction based energy voice activity detection, GMSS/EVAD),希望藉此改善GVAD在低信號雜訊比情況之性能。根據實驗結果,本論文所提的GVAD與GMSS/EVAD方法皆優於G.729、GSM AMR 的VAD模組,並且在方法上也較為簡單。
In speech conversation, there are about 28% of time slices being silent between talk spurts. If a simple encoding method is used to mimic the non-speech, a lot of bit rate will be saved and the computation complexity will be decreased. The purpose of voice activity detection (VAD) is to classify a speech signal into two portions: speech and non-speech.
In most of VAD approaches, an additive signal model is assumed where a noisy speech results from a sum of clean speech and additive noise. Consequently, the performance of VAD heavily depends on noise estimation. Appropriate noise estimation leads to good performance generally. According to estimated noise statistics, a decision rule is constructed for VAD and a determination of speech and non-speech portions is made. In this thesis, we use GM(1,1) model to estimate signal and noise in noisy speech where an additive signal model is assumed. By estimated noise and signal, the segmented signal-to-noise ratio (SNR) is calculated. Based on an adaptive threshold, speech and non-speech segments are determined. The proposed approach is called grey VAD (GVAD).
Besides, we present a grey approach to improve VAD performance for low SNR cases. The proposed approach is called GMSS/EVAD. In the GMSS/EVAD, a GMSS (grey magnitude spectral subtraction) scheme is applied to reduce noise level and to recover speech signal in a noisy speech. In the approach, the GMSS processed speech is considered as a clean-like speech which can be easily classified by an EVAD (energy-based VAD). Consequently, VAD performance may be improved through the GMSS. The results show that proposed GVAD and GMSS/EVAD approaches works well in the examples and has better performance than VAD in G.729 and GSM AMR.
索引目錄
中文摘要 I
ABSTRACT III
誌謝 V
索引目錄 VI
圖目錄 IX
表目錄 XV
第一章 簡介 1
1.1研究背景 1
1.2相關研究 2
1.3 研究動機 7
1.4論文架構 8
第二章 AMR 語音編碼器 9
2.1簡介 9
2.2 AMR之語音活動檢測判斷方式 10
2.2.1濾波器與子波段階層計算 11
2.2.2音高偵測 13
2.2.3音調偵測 14
2.2.4相關複雜訊號分析與偵測 15
2.2.5 VAD的判斷 16
2.2.6 Hangover addition 17
第三章 G.729語音編碼器 19
3.1簡介 19
3.2 G.729之語音活動檢測判斷方式 20
3.2.1語音狀態初步偵測 20
3.2.2 VAD之平滑處理 26
3.2.3四種參數的更新 27
第四章 以GM(1,1)模型為基礎的灰色語音活動檢測方法 29
4.1簡介 29
4.2灰色語音活動檢測方法 29
4.2.1 GM(1,1)模型回顧 31
4.2.2以GM(1,1)模型為基礎的灰色雜訊估計方法 33
4.2.3灰色語音活動檢測流程 39
4.3灰色頻譜削減之能量語音活動檢測方法 41
4.3.1基於灰色信號/雜訊估計之灰色頻譜削減方法 41
4.3.2基於灰色頻譜削減法之能量語音活動檢測方法 44
第五章 模擬結果與討論 46
5.1灰色語音活動檢之模擬結果 46
5.2灰色頻譜削減之能量語音活動檢測方法模擬結果 62
第六章 結論 73
參考文獻 75
[1]L. Karray and A. Martin, “Toward Improving Speech Detection Robustness for Speech Recognition in Adverse Environments,” Speech Communications., no. 3, pp. 261-276, 2003.
[2]Jane-Hwa Huang, Szu-Lin Su, Jiann-Horng Chen, “Design and Performance Analysis for Data Transmission in GSM/GPRS System With Voice Activity Detection,” IEEE Transactions on Vehicular Technology, Vol. 51, no. 4, Jul 2002.
[3]R. Venkatesha Prasad, Abhijeet Sangwan, HS Jamadagni, Chiranth, M.C, Rahul Sah, Vishal Gaurav, “Comparison of Voice Activity Detection Algorithms for VoIP,” International Symposium Computers and Communications , pp. 530-535, Jul 2002 .
[4]J. Sohn, N. S. Kim, and W. Sung, “A Statistical Model-Based Voice Activity Detection,” IEEE Signal Process. Letters, Vol. 16, No. 1, pp. 1-3, Jan. 1999.
[5]J.-H. Chang and N. S. Kim, “Speech Enhancement: New Approaches to Soft Decision,” IEICE Transactions on Information and System, Vol. E84-D, No. 9, Sep 2001.
[6]Yong Duk Cho, Khaldoon Al-Naimi and Ahmet Kondoz, “Improvement Voice Activity Detection Based on Smoothed Statistical Likelihood Ratio,” IEEE Acoustics, Speech, and Signal Processing., Vol. 2, No. 10, pp. 737-740, Oct. 2001.
[7]Joon-Hyuk Chang and Nam Soo Kim, “Voice Activity Detection Based on Complex Laplacian Model,” Electronics Letters, Vol. 39, No. 7, pp. 632-634, Apr 2003.
[8]Javier Ramirez, Jose C. Segura, Carmen Benitez, Angel de la Torre, and Antonio J. Rubio, “A New Kullback-Leibler VAD for Speech Recognition in Noise,” IEEE Signal Processing Letters, Vol. 11, No. 2, pp. 266-269, Feb. 2004.
[9]Javier Ramirez, Jose C. Segura, Carmen Benitez, Angel de la Torre, “Effective Voice Activity Detection Algorithms Using Long-Term Speech Information, ” Speech Communications, Vol. 42, pp. 271-287, 2004.
[10]K.-I. Kim and S.-K. Park, “Voice Activity Detection Algorithm Using Radial Basis Function Network,” Electronic Letters, Vol. 40, No. 22, pp. 1454-1455, Oct. 2004.
[11]H. Misra, S. Ikbal, H. Hermansky, “Spectral Entropy Based Feature for Robust ASR,” International Conference on Acoustics, Speech, and Signal Processing, Vol. I, pp. I-193-196, May 2004.
[12]Li-Ru Lu, “Improved Techniques for Continuous Mandarin Speech Recognition Under Telephone Environment,” Nation Taiwan University, Master Thesis, Jun 1999.
[13]S.-H. Chen and J.-F. Wang, “A Wavelet-based Voice Activity Detection Algorithm in Noise Environments,” International Conference on Electronics, Circuits, and Systems, Vol. 3, pp. 995-998, Sep. 2002.
[14]Voice Activity Detector (VAD) for Adaptive Multi-Rate (AMR) Speech Traffic Channels, ETSI, ETS1 EN 301 708 Recommendation, 1999.
[15]A Silence Compression Scheme for G.729 Optimized for Terminals Conforming to Recommendation V.70, ITU, ITU-T Rec. G.729-Annex B, 1996.
[16]A. Benyassine, E. Shlomot, H. Y. Su, D. Massaloux, C. Lamblin, and J.P. Petit, “ITU Recommendation G.729 Annex B: A Silence Compression Scheme for Use with G.729 Optimized for V.70 Digital Simultaneous Voice and Data Applications,” IEEE Communication. Magazine, vol. 35, pp. 64-73, Sep. 1997.
[17]Min-Chang Chang, “Voice Activity Detection and Its Application to Speech Coding”, Nation Taipei University of Technology Master Thesis, Jan 2003.
[18]Yin-Fan Chen, “Design and Analysis of Multi-Mode Coder With Voice Activity Detection Based on Multiple Linear Regression”, National Taipei University of Technology, Master Thesis, Jun 2005.
[19]J.L. Deng, “Control Problems of Grey System,” System and Control Letters, pp. 288-294, 1982.
[20]J.L. Deng, “Introduction to Grey System Theory,” Journal of Grey System, pp. 1-24, 1989.
[21]K.L. Wen, “ Grey Systems-Modeling and Prediction. Yang''s Scientific Press, Tucson, USA. 2004.
[22] D.G. Childers, Speech Processing and Synthesis Toolboxes, John Wiley & Sons, Inc., 1999.
[23] J. Ramirez, J.C. Segura, C. Benitez, A. de la Torre, and A. Rubio,“An Effective Subband OSF-Based VAD with Noise Reduction for Robust Speech Recognition,”IEEE Transactions on Acoustics, Speech, and Signal Processing, Vol. 13, No.6, pp. 1119-1129, Nov. 2005.
[24] Yasser Ghanbari, Mohannand Reza Karami-Mollaei,“A New Approach for Speech Enhancement Based on The Adaptive Thresholding of Wavelet Packet,”Speech Communications, Vol. 48, pp. 927-940, Aug. 2006.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top
無相關論文
 
系統版面圖檔 系統版面圖檔