對於語詞辨識系統而言,不特定語者(Speaker-Independent)語詞辨識系統實現,具有相當的難度,因此何如建立精確、快速且具強健性的不特定語者辨識系統,一直是一個很大的挑戰。

本研究以隱藏式馬可夫模型(Hidden Markov model ,HMM)為基礎, 建立不特定語者(Speaker-Independent)語詞辨識系統,隱藏式馬可夫模型已經被應用在許多領域, 語音辨識就是其中重要的應用, 隱藏式馬可夫模型以狀態(State)描述語音產生的方式,為一可以代表語音時變特性之統計語音模型。

本研究所採用的語音資料庫分別為OGI(Oregon Graduate Institute of Science and Technology)英文語料及麥克風中文語料, 分別實現特定語者(speaker-dependent)及不特定語者(speaker-independent) 語詞辨識系統。

There are a lot of difficulties that have to be overcome in the speaker-independent (S.I.) phrase recognition system . And the feasibility of accurate ,real-time and robust system pose of the greatest challenges in the system.

In this thesis ,the speaker-independent phase recognition system is based on Hidden Markov Model (HMM). HMM has been proved to be of great value in many applications, notably in speech recognition. HMM is a stochastic approach which characterizes many of the variability in speech signal. It applys the state-of-the-art approach to Automatic Speech Recognition .

目 錄
頁 次
論文摘要………………………………………………………………Ⅰ
致謝……………………………………………………………………Ⅱ
目錄……………………………………………………………………Ⅲ
圖表目錄………………………………………………………………Ⅵ

第一章 緒論...............................................1
1-1 研究動機與目的......................................1
1-2研究方法.............................................2
1-3 論文架構......................................... ..3

第二章 語詞辨識系統與數位語音訊號處理....................5
2-1 語詞辨識系統介紹....................................5
2-2 辨識系統之語音前置處理..............................8
2-2-1 端點偵測(Endpoint Detection)...................10
2-2-2 乘上視窗函數(Window)...........................10
2-3 語音訊號之特徵萃取.................................11
2-3-1倒頻譜(Cepstrum)................................12

第三章 隱藏式馬可夫模型為基礎之語詞辨識系統...........19
3-1 語音訊號之隱藏式馬可夫模型.........................19
3-2隱藏式馬可夫模型之建立..............................20
3-3隱藏式馬可夫模型之訓練..............................21
3-3-1期望值最大演算法 ...............................21
3-3-2參數重估(Parameters Reestimation)...............22
3-4隱藏式馬可夫模型之辨識程序..........................27
3-5 減少搜尋空間解碼演算法.............................30
3-5-1 狀態改變點偵測(Change-Point Point Detection....31
3-5-2 搜尋空間.......................................33

第四章 系統設計實作結果與比較.........................37
4-1 系統設計...........................................37
4-2 系統實作...........................................39
4-3結果與比較..........................................40
4-3-1 語音特徵萃取與辨識率之關係.....................40
4-3-2 語詞訓練次數與辨識率之關係.....................43
4-3-3 模型狀態數與辨識率之關.........................46

第五章 結論與建議................................... .52
5-1 結論...............................................52
5-2 建議...............................................53

參考文獻.................................................54
附錄(一) ...............................................56
附錄(二) ................................................59

參考文獻

[1] Erhan Cinlar, To Stochastic Processes, New Jersey : Prentice
Hall,Inc.,1975.

[2] B. H. Juang and L. R. Rabiner,” Mixture Autoregressive Hidden
Markovmodels for speech signals.” IEEE Trans. Speech and Audio
Processing,vol.33 ,pp 1404-1413, 1985.

[3] J. R. Deller, J. G. Proakis, and J. H. L. Hansen, Discrete Time
Processing of Speech Signals, New York: Macmillan Pub. Co., 1993.


[4] A,M,Kondoz, Digital Speech coding, New York: John Wiley &Sons Inc.,
1994.

[5] R.W. Schafer and J.D. Markel, Eds., Speech Analysis, New York: IEEE
Press, 1979.

[6] J. Makhoul, “Linear prediction: A tutorial review,” Proc. IEEE ,
vol.63 ,pp. 561-580,Feb. 1989.

[7] Lawrence Rabiner and Biing-Hwang Juang, Fundamentals of Speech
Recognition, New Jersey: Prentice Hall,Inc.,1993.

[8] L. R. Rabiner, “A tutorial on hidden Markov models and selected
applications in speech recognition,” Proc. IEEE , vol. 77 , pp.257 -
286 , Feb. 1989.

[9] O. Cappe, C.E. Mokbel, D. Jouvet and E. Moulines, “ An algorithm for
maximum likelihood estimation of hidden Markov models with unknown
state-tying,” IEEE Trans. Speech and Audio Processing,vol. 6 ,pp 61-
70, Jan.1998.

[10] Qi Li, “Search-space reduction for fast , optimal HMM decoding in
speaker verification,” in IEEE Int. Conf. Acoustics, Speech, and Signal
Processing, vol. 2 , pp. II1189 -II1192 , 2000.

[11] Qi Li, “ A detection approach to search-space reduction for HMM state
alignment in speaker verification,” IEEE Trans. Speech and Audio
Processing,vol. 9 ,pp 569-578, July 2001.

[12] M. Bilginer Gulmezoglu, V. Dzhafarov , M. Keskin and A. Barkana, “ A
novel approach to isolated word recognition,” IEEE Trans. Speech and
Audio Processing,vol. 7 ,pp 620-628, Nov. 1999.


[13] Yoshua Bengio,“Markovian Models for Sequential Data,” Neural Computing
Surveys 2,pp.129-162,1999.

[14] 張照煌, “語音辨識技術應用之發展趨勢,”工研院電通所, 民國85年7月.

[15] 龍生雲,“不特定語句之中文語者辨識系統研究,” 國立中山大學電機工程研究所博
士論文, 民國88年11月17日.

[16] 黃俊豪, “大量語者不特定語句環境下語者辨識系統之特徵設計,” 國立中山大學電
機工程研究所碩士論文, 民國90年6月5日.

[17] 黃銘崇, “不特定語者語詞辨識系統之特徵設計,”國立中山大學電機工程研究所碩
士論文, 民國90年6月5日.