Koyama2, Natsuki Ueno1, and Hiroshi Saruwatari1 1The University of Tokyo, Japan 2National Institute of Informatics, Japan Published in IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol, 31, pp. 656-669, 2023. 2023 IEEE International Conference on Acoustics, Speech and Signal Processing
May 1, 2023 3 Synthesizing desired sound field inside with secondary sources (loudspeakers) <latexit sha1_base64="SbdBLMPsjw6ciQkjzJuiJJhjGvk=">AAACF3icbVDLSsNAFJ3UV62vqksXBotQQUoiRV0W3bisYB/QlDCZ3LZDZyZhZiKUkKUf4Te41bU7cevSpX/i9LGwrQcuHM65l3vvCWJGlXacbyu3srq2vpHfLGxt7+zuFfcPmipKJIEGiVgk2wFWwKiAhqaaQTuWgHnAoBUMb8d+6xGkopF40KMYuhz3Be1RgrWR/OJx4qcex3ogeRqCyrKyF/BUZudexKGPz/xiyak4E9jLxJ2REpqh7hd/vDAiCQehCcNKdVwn1t0US00Jg6zgJQpiTIa4Dx1DBeaguunkkcw+NUpo9yJpSmh7ov6dSDFXasQD0zm+WS16Y/E/r5Po3nU3pSJONAgyXdRLmK0je5yKHVIJRLORIZhIam61yQBLTLTJbm5LwDOTibuYwDJpXlTcy0r1vlqq3czSyaMjdILKyEVXqIbuUB01EEFP6AW9ojfr2Xq3PqzPaWvOms0cojlYX7/bXqDc</latexit> udes(r, !) <latexit sha1_base64="OlngG7bMYm0AnUElZFcGXeuuUPg=">AAAB/HicbVA9SwNBEJ2LXzF+RS1tFoNgFe5E1DJoY2cE8wHJEfY2k2TN7t2xuyeEI/4GW63txNb/Yuk/cZNcYRIfDDzem2FmXhALro3rfju5ldW19Y38ZmFre2d3r7h/UNdRohjWWCQi1QyoRsFDrBluBDZjhVQGAhvB8GbiN55QaR6FD2YUoy9pP+Q9zqixUr19J7FPO8WSW3anIMvEy0gJMlQ7xZ92N2KJxNAwQbVueW5s/JQqw5nAcaGdaIwpG9I+tiwNqUTtp9Nrx+TEKl3Si5St0JCp+ncipVLrkQxsp6RmoBe9ifif10pM78pPeRgnBkM2W9RLBDERmbxOulwhM2JkCWWK21sJG1BFmbEBzW0J5Nhm4i0msEzqZ2Xvonx+f16qXGfp5OEIjuEUPLiECtxCFWrA4BFe4BXenGfn3flwPmetOSebOYQ5OF+/In2Vmg==</latexit> ⌦ <latexit sha1_base64="VJ5RMQ2GKmQZdUQJz96dZvLRnxA=">AAAB93icbVA9SwNBEN2LXzF+RS1tFoNgFe4kqGXQxsIiAfMByRH2NnPJkt29Y3dPOI78Alut7cTWn2PpP3GTXGESHww83pthZl4Qc6aN6347hY3Nre2d4m5pb//g8Kh8fNLWUaIotGjEI9UNiAbOJLQMMxy6sQIiAg6dYHI/8zvPoDSL5JNJY/AFGUkWMkqMlZqPg3LFrbpz4HXi5aSCcjQG5Z/+MKKJAGkoJ1r3PDc2fkaUYZTDtNRPNMSETsgIepZKIkD72fzQKb6wyhCHkbIlDZ6rfycyIrRORWA7BTFjverNxP+8XmLCWz9jMk4MSLpYFCYcmwjPvsZDpoAanlpCqGL2VkzHRBFqbDZLWwIxtZl4qwmsk/ZV1buu1pq1Sv0uT6eIztA5ukQeukF19IAaqIUoAvSCXtGbkzrvzofzuWgtOPnMKVqC8/ULSPOTbw==</latexit> L Difficult to solve owing to regional integration Synthesized sound field <latexit sha1_base64="jXBfTdmzlzCxe3eJ2RCk10ZxZxw=">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</latexit> dl <latexit sha1_base64="2143p1RAQMlzka1H5YkQBHUMHYo=">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</latexit> gl(r) • : Driving signal of th secondary sources • : Transfer function of th secondary source <latexit sha1_base64="bKgRyukOmCpfZztJEARaEMlc1nM=">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</latexit> l <latexit sha1_base64="bKgRyukOmCpfZztJEARaEMlc1nM=">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</latexit> l
(PM) – Discretizing target region into control points – Desired pressures are synthesized at the control points – Driving signals are obtained as simple least-squares solution Ø Acoustic Contrast Control (ACC) – Aimed at generating regions of high- and low-acoustic potential energy – Ratio of acoustic potential energy in one region to that in the other region is maximized – Solved as generalized eigenvalue problem May 1, 2023 4 J Simple implementation owing to the closed-form solution L Feasibility depends on the setting of desired phase distribution J Suitable for generating audible and inaudible regions L Power distribution inside the target region cannot be controlled L Flat amplitude response cannot be guaranteed
Ø Optimization problem for PM Ø Optimization problem for AM May 1, 2023 5 Synthesizing desired amplitude (or magnitude) distribution inside the target region, leaving the phase distribution arbitrary <latexit sha1_base64="OlngG7bMYm0AnUElZFcGXeuuUPg=">AAAB/HicbVA9SwNBEJ2LXzF+RS1tFoNgFe5E1DJoY2cE8wHJEfY2k2TN7t2xuyeEI/4GW63txNb/Yuk/cZNcYRIfDDzem2FmXhALro3rfju5ldW19Y38ZmFre2d3r7h/UNdRohjWWCQi1QyoRsFDrBluBDZjhVQGAhvB8GbiN55QaR6FD2YUoy9pP+Q9zqixUr19J7FPO8WSW3anIMvEy0gJMlQ7xZ92N2KJxNAwQbVueW5s/JQqw5nAcaGdaIwpG9I+tiwNqUTtp9Nrx+TEKl3Si5St0JCp+ncipVLrkQxsp6RmoBe9ifif10pM78pPeRgnBkM2W9RLBDERmbxOulwhM2JkCWWK21sJG1BFmbEBzW0J5Nhm4i0msEzqZ2Xvonx+f16qXGfp5OEIjuEUPLiECtxCFWrA4BFe4BXenGfn3flwPmetOSebOYQ5OF+/In2Vmg==</latexit> ⌦ Control points <latexit sha1_base64="lBJueI+9nR6+v444Orp/qYScf/A=">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</latexit> M <latexit sha1_base64="P09u6SpT/CzYrw09JLkb6DzPdBY=">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</latexit> minimize d2CL kGd udesk2 + kdk2 <latexit sha1_base64="xHTP+Y0VPGUaop882+RHZBjsR0M=">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</latexit> minimize d2CL k|Gd| |udes|k2 + kdk2 Driving signals Transfer function matrix Desired pressures Element-wise absolute value
for AM does not have closed-form solution – Proposed algorithm based on alternating direction method of multipliers (ADMM) – Computationally efficient compared with gradient methods Ø Differential-norm penalty for time-domain filter design – Driving signals may have phase discontinuities between frequency bins – To induce smoothness between frequency bins and avoid unnecessarily large time- domain filter length, differential-norm penalty is introduced May 1, 2023 6 Differential-norm penalty Index of frequency bin
Synthesizing desired amplitude distribution over target region, leaving phase distribution arbitrary – ADMM-based algorithm for AM is formulated – Differential-norm penalty for inducing continuities of phase between frequency bins – Desired amplitude distribution is accurately and efficiently synthesized, compared with PM and ACC Ø Code is available: – https://github.com/sh01k/AmplitudeMatching May 1, 2023 9 Paper📝 Code🧑💻 Demo🎥