| Presentation | 2022-03-02 [Poster Presentation] Sound Field Estimation from Small Number of Observations by Deep Learning with Difference-Approximation-Based Helmholtz-Equation Loss Function Kazuhide Shigemi, Shoichi Koyama, TomohikoNakamura, Hiroshi Saruwatari, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | We propose a single-frequency sound field estimation method from a small number of observations that uses a loss function based on the Helmholtz equation for training a convolutional neural network (CNN).Conventional CNN-based sound field estimation methods can enhance the estimation accuracy by using the measurements of the target sound environment. However, since they treat the sound field as a two-dimensional array, their estimated results may be physically infeasible, i.e., those results do not always satisfy the Helmholtz equation. To overcome this problem, we propose a loss function using the difference approximation of the Helmholtz equation, which enables us to encompass the physical constraint of the sound field in the CNN training. Results of numerical experiments show that the proposed method can estimate the sound fields less deviated from the Helmholtz equation while maintaining the accuracy of the sound field estimation. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Sound Field Reconstruction / Physics-Informed Neural Networks / Helmholtz Equation / Finite Difference Approximation |
| Paper # | EA2021-85,SIP2021-112,SP2021-70 |
| Date of Issue | 2022-02-22 (EA, SIP, SP) |
| Conference Information | |
| Committee | EA / SIP / SP / IPSJ-SLP |
|---|---|
| Conference Date | 2022/3/1(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Yoshinobu Kajikawa(Kansai Univ.) / Yukihiro Bandou(NTT) / Norihide Kitaoka(Toyohashi Univ. of Tec) / 北岡 教英(豊橋技科大) |
| Vice Chair | Kenichi Furuya(Oita Univ.) / Shoichi Koyama(Univ. of Tokyo) / Toshihisa Tanaka(Tokyo Univ. Agri.&Tech.) / Takayuki Nakachi(Ryukyu Univ.) |
| Secretary | Kenichi Furuya(NTT) / Shoichi Koyama(RitsumeikanUniv.) / Toshihisa Tanaka(Xiaomi) / Takayuki Nakachi(Takushoku Univ.) / (Tokyo Univ. Agri.&Tech.) / (Univ. of Tokyo) |
| Assistant | Yukou Wakabayashi(Tokyo Metropolitan Univ.) / Tatsuya Komatsu(LINE) / Taichi Yoshida(UEC) / Seisuke Kyochi(Univ. of Kitakyushu) / Toru Nakashika(Univ. of Electro-Comm.) / Ryo Masumura(NTT) |
| Paper Information | |
| Registration To | Technical Committee on Engineering Acoustics / Technical Committee on Signal Processing / Technical Committee on Speech / Special Interest Group on Spoken Language Processing |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | [Poster Presentation] Sound Field Estimation from Small Number of Observations by Deep Learning with Difference-Approximation-Based Helmholtz-Equation Loss Function |
| Sub Title (in English) | |
| Keyword(1) | Sound Field Reconstruction |
| Keyword(2) | Physics-Informed Neural Networks |
| Keyword(3) | Helmholtz Equation |
| Keyword(4) | Finite Difference Approximation |
| 1st Author's Name | Kazuhide Shigemi |
| 1st Author's Affiliation | The University of Tokyo(UTokyo) |
| 2nd Author's Name | Shoichi Koyama |
| 2nd Author's Affiliation | The University of Tokyo(UTokyo) |
| 3rd Author's Name | TomohikoNakamura |
| 3rd Author's Affiliation | The University of Tokyo(UTokyo) |
| 4th Author's Name | Hiroshi Saruwatari |
| 4th Author's Affiliation | The University of Tokyo(UTokyo) |
| Date | 2022-03-02 |
| Paper # | EA2021-85,SIP2021-112,SP2021-70 |
| Volume (vol) | vol.121 |
| Number (no) | EA-383,SIP-384,SP-385 |
| Page | pp.pp.132-139(EA), pp.132-139(SIP), pp.132-139(SP), |
| #Pages | 8 |
| Date of Issue | 2022-02-22 (EA, SIP, SP) |