中文摘要

傳統的噪音消除通常是在單一的通道中演算以增強語音訊號，近年來麥克風陣列的多通道處理技術逐漸被引入噪音消除的觀念之中，在本論文中我們先估算出聲源到達每一個麥克風的延遲時間，利用延遲相加演算(Delay-and-Sum algorithm)，得到一個固定波束形成(Fixed Beamformer)，為了得到一個較好的雜訊與干擾的抑制，我們將經過時間補償後的信號和初步經過固定波束形成的信號送給等適性濾波器(Adaptive Filter)處理，得到一個強健性的等適性波束形成(Robust Adaptive Beamformer)。

我們將等適性波束形成的輸出視為一個單通道的訊號，應用單一通道的語音增強處理，首先我們不以傳統的語音機率估測來得到噪音頻譜，而是利用頻譜上同樣頻率的前一個音框值來決定出區域最小值，並分頻設定不同的臨界參數，直接以一階遞迴演算估算出噪音頻譜，並另外搭配語音存在機率的估測，以OM-LSA (Optimally Modified Log Spectral Amplitude )演算得出一個增益，將等適性波束形成所輸出的信號轉到頻域乘上此增益再以反傅立葉轉換將訊號轉回時域，使用重疊相加演算法(Overlap and add)即可以得到一增強的語音訊號。

參考文獻
[1] Cohen, I.; Berdugo, B.”MICROPHONE ARRAY POST-FILTERING FOR NON-STATIONARY NOISE SUPPRESSION”, Acoustics, speech, and signal processing,2002, Vol 1,pp.901-904
[2] FLANAGAN, JL , JOHNSTON, J D., ZAHN, R., and ELKO, G .W “Computer-steered microphone arrays for sound transduction in large
Rooms”, J Acoust Soc Am , 1985, 78, (5), pp. 1508-1 5 18
[3]SYDOW, C “Broadband beamforming for a microphone array”,
J Acoust Soc Am, 1994,96, (Z), pp. 845-849
[4] GRIFFITHS, J E , and JIM, C W “An alternate approach to linearly
constrained adaptive beamforming”, IEEE Trans Antennas Propag,
1982,30, pp. 27-34
[5] ZUREK, PM , GREENBERG, J E , and PETERSON, PM “Sensitivity
to design parameters in an adaptive-beamforming hearing aid” Proceedings
of ICASSP’90, IEEE international conference on Acoustics, speech,
and signal processing, 1990, Vol. 2, pp. 1129-1 132
[6] Cohen, I.; Berdugo, B.” Noise estimation by minima controlled recursive averaging for robust speech enhancement” IEEE Signal Processing Letters Vol 9, 2002 , pp.12 – 15
[7] Cohen, I.” Noise spectrum estimation in adverse environments: improved minima controlled recursive averaging” IEEE Trans Speech and Aoudio Processing Vol11, 2003, pp.466 – 475
[8] Cohhen, I. “ON SPEECH ENHANCEMENT UNDER SIGNAL PRESENCE UNCERTAINTY”, Acoustics, speech, and signal processing, Vol. 1,2001,pp.661-664
[9] 楊吉文,”以麥克風陣列與語音預估作語音增強之研究” 國立清華大學電機研究所碩士論文，2006
[10] Cohen,I ,”Speech enhancement for non-stationary noise environments” IEEE Trans. Signal Processing 2001, pp.2403-2148
[11] D.Flogeras, R.Doraiswami and M.E.Kaye, “A REAL TIME SPECTRAL SUBTRACTION BASED SPEECH ENHANCEMENT SCHEME” IEEE CCECE Vol 2, 2003, pp.1071 - 1074
[12] D. Malah, R. V. Cox, and A. J. Accardi, “Tracking speech-presence uncertainty to improve speech enhancement in nonstationary noise environments” in Proc. Int. Conf. Acoustics, Speech, Signal Processing, 1999, pp.789-792
[13]M. Omologo, P. Svaizer, “Acoustic event localization using a cross-power spectrum phase based technique” in Proc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing,Vol2, 1994, pp.273-276
[14] M. Omologo, P. Svaizer,”Use of the Cross-power spectrum in acoustic event localization” ISRT Technical Report #9303-13,1993
[15] M. Omologo, P. Svaizer, ”Acoustic source location in noisy and reverberant environment using CSP analysis” IEEE Trans Speech and Audio Processing,Vol2,1996,pp.921-924
[16]陳益正，”使用強健性時間延遲與訊號子空間方法於麥克風陣列語音加強”，國立成功大學資訊工程研究所碩士論文，2003
[17]R.L. Bouquin, A.A. Azirani, and G. Faucon, “Enhancement of Speech Degraded by Coherent and Incoherent Noise Using a Cross-spectral Estimator” IEEE Transactions on Speech and Aoudio Processing, Vol5,No.5,1997,pp.484-487
[18] R.L. Bouquin and G. Faucon, ”Using the Coherent Function for Noise Reduction ” IEE Processings-I, Vol.139,1992,pp.276-280
[19]J. G. Rodriguze et al, “Coherence-based Subband Decomposition for Robust Speech and Speaker Recognition in Noise and Reverberant Room”,IEEE (ICSLP),1998,pp.385-338
[20]Wolfgang, Herbordt ，Springer Verlag “Sound Capture for Human/Machine Interface”,2005,pp.5-23
[21] Wolfgang, Herbordt ，Springer Verlag “Sound Capture for Human/Machine Interface”,2005,pp36-37
[22]王小川編著”語音訊號處理”2005,pp.11-2~11-5
[23]S. F. Bool, “Suppression of acoustic noise in speech using spectral subtraction” in IEEE Trans. Acoustics, Speech and Signal Proc. ,Vol. 27,1979, pp. 113-120
[24]Y. Ephraim and D. Malah, “Speech enhancement using a minimum mean-square error log-spectral amplitude estimator” in IEEE Trans. Acoustics , Speech, Signal Processing, Vol. ASSP-33 ,1985,pp.443-445
[25] Y. Ephraim and D. Malah, “Speech enhancement using a minimum mean-square error short-time spectral amplitude estimator” in IEEE Trans. Acoustics , Speech, Signal Processing, Vol. ASSP-32, 1984, pp.1109-1121
[26]Martin, R.” Noise power spectral density estimation based on optimal smoothing and minimum statistics” IEEE Trans. Speech Aoudio Processing, Vol9 ,2001,pp.504-512