傳統Fast ICA的獨立成份分析(Independent Component Analysis, ICA)方法會面臨兩種缺點，一個是獨立成份的順序無法決定，另一個是Fast ICA也僅限於解決靜態資料。為了要避免這樣的問題，本研究提出一種基於Robust ICA與頻域訊號分離改進的方法。Robust ICA是一種利用四次多項式的根來尋找分離矩陣，達到收斂效果以求得獨立成份，提供了複數運算的支援並減少白化的過程。將ICA的操作過程放到頻域中執行可以將龐大的問題分成好幾個子問題個別處理，利用硬體平行處理的特性同時分離好幾個聲音段可以加快語音訊號分離處理速度，另外一個好處是在頻域中非高斯的情況會比時域中更加明顯。實驗結果可以發現，基於此架構設計出的語音訊號分離硬體加速器可以解決語音訊號順序不確定的問題，達到語音訊號批次處理的效果。在語音訊號執行速度上，語音訊號分離的處理速度比軟體快，並可以解決實驗中無法達到的語音即時處理的效果。

The traditional independent component analysis of FastICA faces two of disadvantages, undetermined component of order and offline experiment. We proposed a new solution by using RobustICA algorithem with demixing signal in the frequency domain. RobustICA separate independent component by searching demixing vector and using four degree polynomial, supporting complex calculation without whitening progress. The progress which separated component in frequency domain will speed up separated by divided a bunch data into frequency bin with hardware characteristic of parallel process. Moreover, the Non-Gaussianity is obviously in frequency domain. Its shows in experiment that the voice signal separated accelerator have many characteristics including batching progress, ordering component and real time signal separated process.

[1] Pleva, M. ; Ondas, S. ; Juhar, J. ; Cizmar, A. ;Papaj, J. ; Dobos, L, "Speech and mobile technologies forcognitive communication and informationsystems," 2011 2nd International Conference on Cognitive Infocommunications, pp. 1-5, 2011.

[2] Siri , [Online]. Avaliable: http://en.wikipedia.org/wiki/Siri on 30 June 2014.

[3] Google Voice Search, [Online]. Avaliable: http://en.wikipedia.org/wiki/Google_Voice_Search on 2 January 2014.

[4] S Voice , [Online]. Avaliable: http://en.wikipedia.org/wiki/S_Voice on 17 June 2014.

[5] Tomar, V.S. ; Rose, R.C, "A Family of Discriminative Manifold Learning Algorithms and Their Application to Speech Recognition," IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 22, no. 1, pp. 161-171, 2014.

[6] Narayanan, A. ; DeLiang Wang, "Investigation of Speech Separation as a Front-End for Noise Robust Speech Recognition," IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol.22, no. 4, pp. 826-835, 2014.

[7] Daphal, S.D. ; Jagtap, S.K, "Noise Robust Novel Approach to Speech Recognition," 2014 International Conference on Electronic Systems, Signal Processing and Computing Technologies (ICESC), pp. 289-294, 2014.

[8] B. Van Veen and K. Buckley, “Beamforming: a versatile approach to spatial filtering,” ASSP Magazine, IEEE, vol. 5, no. 2,pp. 4 –24, april 1988.

[9] Hidri Adel, Meddeb Souad, Abdulqadir Alaqeeli, Amiri Hamid, "Beamforming Techniques for Multichannel audio Signal Separation", JDCTA: International Journal of Digital Content Technology and its Applications, Vol. 6, No. 20, pp. 659-667, 2012.

[10] Thiemann, J. ; Vincent, E, "An experimental comparison of source separation and beamforming techniques for microphone array signal enhancement," 2013 IEEE International Workshop on Machine Learning for Signal Processing (MLSP), pp. 1-5, 2013.

[11] J. H´erault, C. Jutten, and B. Ans, 「D´etection de grandeurs primitives dans un message composite par une architecture de calcul neuromim´etique en apprentissage non supervis´e,」 in Proc. GRETSI, Nice, France, 1985, pp. 1017–1020

[12] Ouedraogo, W.S.B. ; Souloumiac, A. ;Jaidane, M. ; Jutten, C, "A robust geometrical method for blind separation of noisy mixtures of non-negatives sources," 2013 8th International Workshop on Systems, Signal Processing and their Applications (WoSSPA), pp. 37-41, 2013.

[13] Even, J. ; Saruwatari, H. ; Shikano, K, "Frequency domain semi-blind signal separation: application to the rejection of internal noises," Conference on Acoustics, Speech and Signal Processing, 2008. ICASSP 2008. IEEE International, pp. 157-160, 2008.

[14] Hongjuan Sun ; Qi Zhu, "Blind speech signal separation in wireless sensor networks," vol. 03, pp. 1422-1426, 2013.

[15] A. Hyvarinen, J. Karhunen and E. Oja, Independent Component Analysis. John Wiley & Sons, Inc., New York, 2001.

[16] Hyvärinen, A.; Oja, E, "Independent component analysis: Algorithms and applications". Neural Networks 13 (4–5): 411–430, 2000.

[17] T.W. Lee, “Independent Component Analysis: Theory and Applications,” Kluwer Academic Publishers, Boston, 1998.

[18] J.-F. Cardoso and A. Souloumiac, ‘‘Blind heamforning for non-Gaussian signals,” in Pmc. IEE -F,vol. 140, pp. 362-370, Dec. 1993.

[19] Lu, W.; Rajapakse, JC. Constrained independent component analysis. In: Leen, TK.; Dietterich,TG.; Tresp, V., editors, “Advances in Neural Information Processing Systems,” Vol. 16, pp. 570-576, 2000.

[20] Ze Wang, "Fixed-point algorithms for constrained ICA and their applications in fMRI data analysis," Magn Reson Imaging, November 201.

[21] Li-Yuan Chen and Chi-Jie Lu, Member, IACSIT, "An Improved Independent Component Analysis Algorithm Based on Artificial Immune System" International Journal of Machine Learning and Computing, Vol. 3, No. 1, February 2013.

[22] Zarzoso, V. ; Comon, P, "Robust Independent Component Analysis by Iterative Maximization of the Kurtosis Contrast With Algebraic Optimal Step Size," IEEE Transactions on Neural Networks, Vol. 21, No. 2, pp.248-261, 2010.

[23] Zarzoso, V. ; Comon, P, “Comparative speed analysis of Fast ICA,”in Proc. ICA-2007, 7th Intl. Conf. Independent Component Analysis and Signal Separation, London, UK, pp. 293–300, Sept. 9–12, 2007.

[24] Okamoto, R. ; Takahashi, Yu. ; Saruwatari, H.; Shikano, K, "MMSE STSA estimator with non stationary noise estimation based on ICA for high-quality speech enhancement," 2010 IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), pp. 4778-4781, 2010.

[25] Wanlong Li ; Ju Liu ; Jun Du ; Shuzhong Bai, "Microphone array speech enhancement system combining ICA preprocessing in highly noisy environments," International Conference on Audio, Language and Image Processing, 2008. ICALIP 2008, pp. 649-652, 2008.

[26] Esaki, S. ; Niwa, K. ; Nishino, T. ; Takeda, K, "Estimating sound source depth using asmall-size array," 2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 401-404, 2012.

[27] Lombard, A. ; Yuanhang Zheng ;Kellermann, W, "Synthesis of ICA-based methods for localization of multiple broadband sound sources," 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 157-160, 2011.

[28] Ogasawara, M. ; Nishino, T. ; Takeda, K, "A small dodecahedral microphone array forblind source separation," 2010 IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), pp. 229-232, 2010.

[29] Noohi, T. ; Epain, N. ; Jin, C.T, "Direction of arrival estimation for spherical microphone arrays by combination of independent component analysis and sparse recovery," 2013 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 346-349, 2013.

[30] Nesta, F. ; Omologo, M, "Cooperative Wiener-ICA for source localization and Separation by distributed microphone arrays,"2010 IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), pp. 1-4, 2010.

[31] E. B. Hogenauer, “An economical class of digital filters for decimation and interpolation,” IEEE Transactions on Acoustics, Speech, and Signal Processing, vol. ASSP-29, No.2, pp.155-162, April 1981.
[32] Pulse-density modulation, [Online]. Avaliable: http://en.wikipedia.org/wiki/Pulse-density_modulation. on 12 June 2014.
[33] Fredric J. Harris ”Multirate Signal Processing for Communication Systems”.
[34] Aapo Hyvärinen, Juha Karhunen, Erkki Oja, “Independent Component Analysis”
[35] A. Rinen,“New approximations of differential entropy for independent omponent analysis and projection pursuit” In Advances in Neural InformationProcessing Systems, vol 10, pp 273-279. MIT press, 2003.
[36] W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery, Numerical Recipes in C. The Art of Scientific Computing, 2nd ed. Cambridge, U.K.: Cambridge Univ. Press, 1992.
[37] S. Makino, H. Sawada, R. Mukai, S. Araki, “Blind Source Separation of Convolutive Mixtures of Speech in Frequency Domain,” IEICE Trans. Fundamentals, vol. E88-A, no. 7, pp. 1640-1655, 2005.
[38] M. S. Pedersen, J. Larsen, U. Kjems, L. C. Parra, “A Survey of Convolutive Blind Source Separation Methods,” Springer Handbook on Speech Processing and Speech Communication, 2007.
[39] N. Mitianoudis and M. E. Davies, “Audio Source Separation of Convolutive Mixtures,” IEEE Transactions on Speech and Audio Processing, vol. 11, no. 5, pp. 489-497, 2003.
[40] Ching-Han Chen , Yao TK, Dai JH and Chen-Yuan Chen, “A pipelined multiprocessor SOC design methodology for streaming signal processing”. Journal of Vibration and Control, February 2014, vol. 20 no. 2, 163-178.
[41] Ching-Han Chen , Chia-Ming Kuo, Chen-Yuan Chen, and Jia-Hong Dai, “The design and synthesis using hierarchical robotic discrete-event modeling”, Journal of Vibration and Control, December 2013, vol.19, no.11, pp.1603–1613.