臺灣博碩士論文加值系統

3D虛擬音效之模擬是一研究層面相當廣的主題，其中包含了決定虛擬聲源與聆聽者之關係的定位問題，以及提供空間感受的殘響音效問題。本論文將側重於後者之研究，藉由將專屬於某種空間之反射與殘響的重新合成，以還原於此空間聆聽聲音的感受。然而，由於空間的模態極端複雜，使得空間響應的模型化產生困難，因而需要超長階數的摺積動作才能重現此一聲場。基於一已量測的房間響應為設計樣版，我們提出兩種可行的方式來滿足即時實現殘響音效的需求，第一種方法為在餘弦調變之濾波器組進行平行化之次頻帶濾波，並且將次頻帶的濾波器以IIR 型式的結構做模型化的處理，而得到進一步的運算量精簡；第二種方式則考慮人耳對不同頻帶有不同的解析度的特性，配合非等間隔採樣，用一階數大幅減少的FIR濾波器來取代原本所需的超長階數摺積。這些方法均已證明確實有效，與既存的常用方法比較，不僅可提供較自然的音效而且沒有不悅耳的聲音缺陷問題。

Reproducing the 3D virtual sound effect is a broad topic involving the positioning information which provides the relationship between sound sources and listeners, and the room effect which gives the spaciousness perception. This research aims for regenerating the environmental context via synthesizing the reflections and reverberations pertaining to particular listening space. However, the modal distribution of the room response is very complicated, leading to the difficulty of extremely long convolution. Given a measured room response, we propose two applicable schemes to reproduce a reverberant environment with real-time performance. First, the subband filtering conducted in cosine modulated filter bank is used to implement the reverberator in parallelism with subband filters modeled as IIR-based structures. Second, the non-uniformly sampling approach considering the frequency dependent resolution of the human listening system is used to replace the long convolution filter with a much fewer taps FIR filter. These algorithms are validated to attain a natural-sounding room effect without the rendering deficiency of commonly used reverberators.

TABLE OF CONTENTS
LIST OF TABLES……………………………………………………i
LIST OF FIGURES…………………………………………………ii
CHAPTER 1 INTRODUCTION…………………………………………………1
1.1 Motivation ………………………………………………………….1
1.2 Scope of This Thesis………………………………………………2
1.3 Organization…………………………………………………………3
CHAPTER 2 REPRODUCTION OF 3D SOUND EFFECT……………………….5
2.1 Psychoacoustics of Spatial Hearing……………………………6
2.1.1 Azimuth Cues………………………………………………………6
2.1.2 Elevation Cues……………………………………………………9
2.1.3 Range Cues…………………………………………………………10
2.1.4 Head-Related Transfer Function………………………………11
2.1.5 Headphones VS Loudspeakers……………………………………12
2.2 Room Response……………………………………………………….13
2.2.1 Modeling Early Reverberation…………………………………15
2.2.2 Modeling Late Reverberation………………………………….17
CHAPTER 3 SUBBAND FILTERING IN COSINE MODULATED FILTER BANK.21
3.1 Cosine Modulated Filter Bank……………………………………23
3.1.1 Cosine Modulated Perfect Reconstruction QMF…………….24
3.1.2 Cosine Modulated Pseudo QMF………………………………….26
3.2 Filtering in Subband Domain…………………………………….28
3.3 Subband Filter structures for Room Response……………….30
3.3.1 IIR Filter Identified in Frequency Domain……………….31
3.3.2 Parameterization with FIR Filter and a Comb filter……31
3.3.3 Comb Filters and Nested Allpass Filters Optimized via GA……………………………………………………………………………34
3.3.4 Comb Filters and Nested Allpass Filters………………….40
CHAPTER 4 NON-UNIFORMLY SAMPLING SCHEME………………………….42
4.1 Critical Bands………………………………………………………44
4.2 The Non-uniformly Sampling Scheme…………………………….45
4.3 Synthesis of Optimal Non-uniform FIR Filters………………47
4.4 Applications in 3D Sound Reproduction……………………….50
4.4.1 3D sound positioning with HRTF………………………………50
4.4.2 Synthesis of room response……………………………………52
CHAPTER 5 CONCLUSIONS………………………………………………….56
REFERENCES…………………………………………………………………61
LIST OF TABLES
TABLE 1. Coefficients of the Comb Filters for the Second Model in the Subband Filtering Scheme…………………………………….65
TABLE 2. Coefficients of the Comb and Nesting Allpass Filters in the Fifth Subband for the Second Model……………………….65
TABLE 3. Comparison of Computational Load for the Three Models Used in the Subband Filtering Scheme………………………………66
TABLE 4. Center Frequencies and Bandwidths of Critical Bands67
TABLE 5. Comparison of Computational Load in the Proposed Methods for Room Effect Reproduction………………………………68
LIST OF FIGURES
Figure 1. A model for implementation of spatial hearing…….69
Figure 2. Highly simplified, schematic overview of the auditory system………………………………………………………….69
Figure 3. Commonly used spherical coordinate systems for specify the location of a sound source relative to the listener……………………………………………………………………70
Figure 4. Illustration of the azimuth cue, ITD, with sound source locating on horizontal plane……………………………………………………………………….70
Figure 5. Perceptual effect of increasing ITD from 0 to 40 msecs……………………………………………………………………….71
Figure 6. Frequency responses for two different directions of arrival used to explain the elevation cue provided by pinna……………………………………………………………………….71
Figure 7. Crosstalk problem from using loudspeaker to reproduce virtual sound……………………………………………….72
Figure 8. The ideal impulse response of a room…………………………………………………………………………72
Figure 9. Energy decay curve of a room response for one subband…………………………………………………………………….73
Figure 10. Sound pressure level in a room, versus log of distance to sound source………………………………………………73
Figure 11. Early reverberation models. (a) Combining early echoes and the late reverberation. (b) FIR filter cascaded with reverberator……………………………………………………….74
Figure 11. Early reverberation models. (c) Average head-related filter applied to a set of early echoes……………….75
Figure 12. Late reverberation models. ……………………………76
Figure 13. (a) Stautner and Puckette’s four channel FDN. (b) FDN as a general specification of a reverberator containing N delays [Jot and Chaigne]………………………………………………77
Figure 14. Representation of the power complementary pair of functions using a lossless lattice…………………………………………………………………….78
Figure 15. Cosine modulated filter bank with perfect reconstruction. Figure 35. Cosine modulated pseudo QMF filter bank.……………………………………………………………………….79
Figure 16. Cosine modulated pseudo QMF filter bank…………………………………………………………………………80
Figure 17. Comparison of direct convolution and subband filtering.…………………………………………………………………81
Figure 18. Block diagrams of filtering in two domains.………82
Figure 19. Subband filtering scheme. …………………………….83
Figure 20. Illustration of subband filtering in Fig. 37 with polyphase components……………………………………………………84
Figure 21. Spectrum contents of the signals in 3-channel filter bank……………………………………………………………….85
Figure 22. Measured room response of a concert hall………….86
Figure 23. Implementation results of subband filtering with the first structure…………………………………………………….87
Figure 24. Implementation results of subband filtering with the second structure……………………………………………………88
Figure 25. The third structure used for subband filtering, consisting of parallel comb filters and nested allpass filters…………………………………………………………………….89
Figure 26. Comparison of impulse responses of (a) two allpass filters in series and (b) 2-loop nested allpass filter………………………………………………………………………90
Figure 27. The relationship between the fitness function and the cost function……………………………………………………….91
Figure 28. Flow chart for illustration of the procedures in GA……………………………………………………………………………92
Figure 29. Implementation results of subband filtering with the third model. (a) Comparison of energy decay curves in the fifth subband. (b) Comparison of impulse responses in the fifth subband.……………………………………………………………93
Figure 29. Implementation results of subband filtering with the third model. (c) Comparison of frequency responses in the fifth subband.……………………………………………………………94
Figure 30. Listening test outcomes to compare the three subband modeling structures.…………………………………………95
Figure 31. Implementation results of using parallel comb filters cascaded with nested allpass filter………………………………………………………………………96
Figure 32. The critical band distribution………………………………………………………………97
Figure 33. Comparison of low-pass filter responses designed using the non-uniform sampling approach with various norm criteria. (a) The optimized filter via norm, (b) the optimized filter via norm………………………………………….98
Figure 33. Comparison of low-pass filter responses designed using the non-uniform sampling approach with various norm criteria. (c) the optimized filter via norm.……………………99
Figure 34. The HRTF’s at zero azimuth and zero elevation…100
Figure 35. Comparison of HRTF responses obtained using the non-uniform and uniform filters for the direction indicating the azimuth and elevation angles, respectively…………………………………………………………….101
Figure 36. Comparison of HRTF responses obtained using the non-uniform and uniform filters for the direction indicating the azimuth and elevation angles, respectively…………………………………………………………….102
Figure 37. Subjective listening test of sound localization. (a) Instruction of how to interpret the box plot……………………………………………………………………….103
Figure 37. Subjective listening test of sound localization. (b) The result on the azimuth plane. (c) The result on the elevation plane…………………………………………………………104
Figure 38. The Moorer’s reverberator with .comb, all-pass, , , and represent the comb filter, all-pass filter, gain, pure delay, and low-pass filter, respectively…………………………………………………………….105
Figure 39. Comparison of the room responses obtained using the non-uniform sampling technique and direct convolution………………………………………………………………106
Figure 40. Subjective listening test for reverberation effect. (a) Symphony music. (b) Female speech…………………………………………………………………….107

REFERENCES
[1] Zakarauskas, P., and Cynader, M. S., “A computational theory of spectral cue localization,” J. Acoust. Soc. Amer., 94, 1323-1331, 1993.
[2] Sabine, W. C., “Reverberation,” In Lindsay, R. B., editor, Acoustics: Historical and Philosophical Development. Dowden, Hutchinson, and Ross, Stroudsburg, PA.
[3] Schroeder, M. R., “New method of measuring reverberation time,” J. Acoust. Soc. Amer., 37, 409-412, 1965.
[4] Begault, D. R., “Perceptual effects of synthetic reverberation on three-dimensional audio systems,” J. Audio Eng. Soc. 40, 895-904, 1992.
[5] Schroeder, M. R., “Digital simulation of sound transmission in reverberant spaces,” J. Acoust. Soc. Amer., 47, 424-431, 1970.
[6] Moorer, J. A., Chauveau, A., Abbott, C., Eastty, P., and Lawson, J., The 4C Machine. Computer Music Journal, 3, 244-250, 1979.
[7] Jot, J. M., Larcher, V., and Warusfel, O., “Digital signal processing issues in the context of binaural and transaural stereophony,” In Proc. Audio Eng. Soc. Conv. Preprint 3980, 1995.
[8] Schroeder, M. R., “Natural Sounding Artificial Reverberation,” J. Audio Eng. Soc., 10(3).
[9] Stautner, J. and Puckette, M., “Designing multi-channel reverberators,” Computer Music Journal, 6, 52-65.
[10] Jot, J. M., Etude et realization d’un spatialisateur de sons par modeles physiques et perceptifs (Design and implementation of a sound spatializer based on physical and perceptual models, in French). PhD thesis, Telecom Paris.
[11] Jot J. M., “An analysis/synthesis approach to real-time artificial reverberation,” Proc. IEEE Int. Conf. Acoust., Speech and Signal Proc., vol. 2, pp. 221-224, 1992.
[12] Moore, J. A., “The synthesis of complex audio spectra by means of discrete summation formulae,” J. Audio Eng. Soc., vol. 24, pp. 717-727, 1976.
[13] Moore, J. A., “About the reverberation business,” Computer Music Journal, vol. 3, pp. 3255-3264, 1979.
[14] Lake Technology Ltd., http://www.lake.com.au
[15] Gardner, W. G., “Efficient convolution without input-output delay,” J. Audio Eng. Soc., vol. 43, pp. 127-136, Mar. 1995.
[16] Bosi, M., “Filter banks in perceptual audio coding,” AES 17th Conference on high-quality audio coding, Sep. 2-5, 1999, Florence, Italy.
[17] P. P. Vaidyanathan, “Orthonormal and biorthonormal filter banks as convolvers, and convolutional coding gain,” IEEE Tran. on SP, pp. 2110-2130, June 1993.
[18] M. Vetterli, “Running FIR and IIR filtering using multirate filter banks,” IEEE Trans. on ASSP, pp. 730-738, May 1988.
[19] C. A. Lanciani, and R. W. Schafer, “Subband-domain filtering of MPEG audio signals,” IEEE Int. Conf. on Acoust., Speech, Signal Proc., 1999, Arizona, USA.
[20] J. R. VandeKieft, “Computational Improvements to Linear Convolution With Multi-rate Filtering Methods,” Master thesis, Univ. of Miami, Florida, USA, 1998.
[21] P. P. Vaidyanathan, Multirate Systems and Filter banks. Englewood Cliffs, N. J., PTR Prentice Hall, 1993.
[22] Y. P. Lin, and P. P. Vaidyanathan, “A Kaiser window approach to the design of prototype filters of cosine modulated filter banks,” IEEE Signal Processing Letters, pp. 132-134, June 1998.
[23] Physikalisch-Technische Bundesanstalt, project 1.401: simulation of room acoustics, http://www.ptb.de/en/org/1/14/1401/_index.htm
[24] U. Zolzer, Digital Audio Signal Processing. N. Y., John Wiley and Sons, 1997.
[25] W. G. Gardner, “A realtime multichannel room simulator,” J. Audio Eng. Soc. Am., 92(A), 2395, 1992.
[26] B. S. Chen and Y. M. Cheng, “A structure-specified optimal control design for practical applications: a genetic application,” IEEE trans. on control systems technology, vol. 6, No. 6, Nov. 1998.
[27] T. Kozek, T. Roska, and L. O. Chua, “Genetic algorithm for CNN template learning ,” IEEE Trans. Circuits Syst., vol. 40, pp. 392-402, 1993.
[28] S. K. Mitra, S. Chakrabarti, and E. Abreu, “Nonuniform discrete Fourier transform and its application in signal processing,” Proc. EUSIPCO, vol. 2, pp. 909-912 (1992).
[29] J. M. Youngberg and S. F. Boll, “Constant-Q signal analysis and synthesis,” Proc. ICASSP, pp. 375-378 (1978 April).
[30] J. M. Kates, “Constant-Q analysis using the chirp z-transform,” Proc. ICASSP, CH-1379, pp. 314-317 (1978).
[31] J. Garas and P. C. W. Sommen, “Real-time convolution and correlation using no-uniform spectral analysis and synthesis,” Proc. Signal Processing Symposium SPS 98, pp. 95-98 (1998 March).
[32] J. Garas and P. C. W. Sommen, “Warped linear time invariant systems and their application in audio signal processing,” Proc. ICASSP 99, IEEE International Conference on Acoustics, Speech, and Signal Processing, vol.3, pp. 1521-1524 (1999 March).
[33] L. Cohen, Time-frequency analysis, (Englewood Cliffs, NJ, Prentice-Hall, 1995).
[34] J. J. Clark, M. R. Palmer, and P. D. Lawrence, “A transformation method for the reconstruction of functions from nonuniformly spaced samples,” IEEE trans. ASSP, vol. 33, pp. 1151-1165 (1985 October).
[35] S. Boyd, L. Vandenberghe, and M.Grant, “Efficient convex optimization for engineering design,” Proc. IFAC Symp. Robust Contr. Design. (Rio de Janeiro, Brazil, 1994 Sept.).
[36] E. Zwicker and H. Fastl, Psychoacoustics: facts and models. (Berlin Heidelberg, Springer-Verlag, 1999).
[37] B. Noble, Applied linear algebra. (Prentice Hall, Englewood Cliffs, NJ, 1969).
[38] B. Gardner and K. Martin, “HRTF measurement of a KEMAR dummy-head microphone,” MIT Media Lab Perceptual computing-technical report #280 (1994).
[39] J. Chambers, W. Cleveland, B. Kleiner, and P. Tukey, Graphical methods for data analysis. (Belmont: Wadsworth, 1983).
[40] T. P. Krauss, L. Shure, J. N. Little, Matlab Signal Processing Toolbox, The Math works, Inc., 1993.
[41] A. Grace, Matlab Optimization Toolbox, The Math works, Inc., 1995.