本論文的研究目標，是建立應用在晶圓塗料過程中的異常音辨識
系統。論文著重於建立不易受到環境音干擾的異常音辨識系統。主要研究在於前端環境音處理，以及利用異常聲音頻譜之區別所設計的帶通濾波器組所建立的聲音特徵值。
在識別系統的前端，我們透過頻譜刪減法(spectral subtraction)
來消除環境音，而針對其殘餘誤差能量的問題，我們提出平滑地學習環境音的方式，以及利用在一音框內頻譜的變化量，來消除誤差能量，並增加頻譜分辨能力。
在聲音特徵值的取得，是透過一組帶通濾波器。利用異常音的頻
率特性不同，而設定出不同的帶通濾波器組中各頻帶中心頻率及頻帶寬度。經實驗結果驗證，按照本篇論文的方法所設計的帶通濾波器組，具有較強的抗雜訊干擾能力，且較原有方法更加節省資源，程式運算方式更為簡單。
透過實驗結果，結合改良的spectral subtraction 及利用聲音特性所建立的帶通濾波器組，能有效的抑制寬帶的環境噪音，且識別異常聲音。而在實驗中，我們也對異常音加入訊噪比 4db 至-5db 的
高斯寬帶噪音，辨識正確率均能保持在90%以上。

The research target of this thesis is to establish an abnormal voice recognition system in a wafer-coating environment; The thesis focuses on setting up the abnormal voice recognition system which will not be interfered easily with by the sounds of the environment. The main research includes the management of the environmental voice
in the beginning stage, and the voice feature established by the band-pass filters which is designed adapting to the abnormal voice differences.
On the beginning stage of the recognition system, we intend to reduce the environmental sound by spectral subtraction. Regarding to the remaining spectral power problem, we propose the way of a smooth learning environment sound, and utilizing the spectral change amount in a sound frame to reduce spectral power, in the
meantime increasing the spectral resolution capability.
The voice feature is obtained by a band-pass filters. We adapt the setting of every frequency band center and bandwidth in the band-pass filters by the differences in
frequency characteristics of the abnormal voice. According to the experimental result,band-pass filters designed by the method of this thesis have better results resisting the
interference of other undesired signals. It uses less resources and easier calculation than the existing method.
Through the experimental result, combine spectral subtraction improved and with the band-pass filters which are designed adapting to the abnormal voice differences, can be effective inhibition the environment voice and broad-band noise, and recognition the abnormal voice. At experiment, when we join Gaussian of broadband noises of
SNR 4dB to -5dB to make uproar to the abnormal voice, the correct rate of distinguish abnormal voice can keep above 90%.

[1] A. Harma, “Automatic Identification of Bird Species Based on Sinusoidal Modeling of Syllables,” in Proceedings of International Conference Acoustics,
Speech, and Signal Processing, Vol. 5, pp. 545-548, 2003.
[2] B. W. Zhang, “The Study on Corpus-Based Analysis for Bird Sound Recognition System,” Master Thesis, National Dong Hwa University, July 2003.
[3] C. C. Hsu, “Audio Signal Classification”, Master Thesis, National Tsing Hua University, July 2001.
[4] E. D. Chesmore, “Application of Time Domain Signal Coding and Artificial Neural Networks to Passive Acoustical Identification of Animals,” Applied Acoustics,
Vol. 62, No. 12, pp. 1359-1374, December 2001.
[6]王小川， 語音訊號處理， 全華科技圖書， 2005
[7] D. Li, I. K. Sethi, N. Dimitrova, and T. McGee, “Classification of General Audio Data for Content-Based Retrieval,” Pattern Recognition Letters, Vol. 22, No.5,
pp. 533-544, April 2001.
[8] T. Zhang and C.-C. J. Kuo, “Audio Content Analysis for Online Audiovisual Data Segmentation and classification,” IEEE Trans. on Speech and Audio
Processing, Vol. 9, No. 4, pp. 441-457, May 2001.
[9] G. J. Lu, “Indexing and Retrieval of Audio: A Survey,” Multimedia Tools and Applications, Vol. 15, pp. 269-290, 2001.
[10] E. Wold, T. Blum, D. Keislar, and J. Wheaten, “Content-Based Classification,Search, and Retrieval of Audio,” IEEE Multimedia Magazine, Vol. 3, No. 3, pp.
27-36, Fall 1996.
[11] Rabiner L R ,Schafer R W. Digital Processing of Speech Signals, Englewood Cliffs :Prentice Hall ,1978.
[12] Ross M ,Shaffer H, Cohen A, et al. “Average magnitude difference function pitch extractor,” IEEE Trans on Acoustics , Speech ,and Signal Processing, 22 (5) , pp.353 – 362, 1974.
[13] R. Vergin, D. O'Shaughnessy, and A. Farhat, “Generalized Mel Frequency Cepstral Coefficients for Large-Vocabulary Speaker-Independent Continuous-Speech Recognition,” IEEE Trans. on Speech and Audio Processing,
Vol. 7, No. 5, pp. 525-532, September 1999.
[14] R. Vergin, D. O’Shaughnessy, and V. Gupta, “Compensated Mel Frequency Cepstrum Coefficients,” in Proceedings of International Conference Acoustics,
Speech, and Signal Processing, pp. 323-326, May 1996.
[15]林威延， 基於聽覺特性之LPC編碼之研究， 國立中原大學電機工程學系碩士論文，2003。
[16] A. Acero, R. M. Stern, “Cepstral Normalization for Robust Speech Recognition”,Proc. of Speech Processing in Adverse Conditions, pp. 89-92, Cannes-Mandelieu,France, 1992.
[17] H. Hermansky, “Perceptual Linear Predictive (PLP) Analysis of Speech”, Journal Acoust. Soc. Am., Vol. 87, No. 4, pp. 1738-1752, 1990.
[18]陳松琳， 以類神經網路為架構之語音辨識系統， 國中中山大學電機工程學系碩士論文
[19] A. Betkowska, K. Shinoda, and S. Furui, ”Robust speech recognition using factorial HMMs for home environments,” Eurasip Journal on Applied Signal
Processing, in press, 2007.
[20] A. Rosenberg, C.-H. Lee, F. Soong, “Cepstral Channel Normalization Techniques for HMM-Based Speaker Verification”, Proc. of ICSLP’94, pp.1835-1838, Yokohama, Japan, 1994.
[21] B. Logan and P. Moreno, “Factorial HMMs for Acoustic Modeling,” in Proc.ICASSP, pp. 813-816, 1998.
[26] Boll S. “Suppression of Acoustic Noise in Speech using Spectral Subtraction,”IEEE Trans. On Acoustics, Speech and Signal Processing, 27(2) , pp. 113-120 , 1979.
[27]Van Compernolle D. “Noise Adaptation in a Hidden Markov Model Speech Recognition System.,” Computer Speech and Language, 2(2) , pp. 151-167, 1988.
[28]Berouti M, Schwartz, R, Makhoul J. “Enhancement of Speech Corrupted by Acoustic Noise,” Proceeding of 1979 IEEE ICASSP, pp. 208-211, 1979.
[29] M.J.F.Gales et al., “Cepstral parameter compensation for HMM recognition in noise”, Speech Communication,Vol.12, No.3, pp. 231-239, 1993.
[30]J. L. Gauvain et al., “Maximum a posteriori estimation for multivariate Gaussian mixture observations of Markov chains”, IEEE Transactions on Speech and Audio
Processing, Vol. 2, No. 2, pp. 291-298 , 1994.
[31]Haimi-Cohen, R.、 Rannon, Z.M., “Dynamic Time Warping with Generalized Templates for Speaker Independent Speech Recognition,” IEEE Conference，Electrical and Electronics Engineers in Israel, pp. 1549-1552, 1989.