Presentation | 2022-07-13 Investigation of noise removal using U-Net and voice recognition performance improvement Jian Lin, Shota Sano, Yuusuke Kawakita, Tsuyoshi Miyazaki, Hiroshi Tanaka, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | A method for converting noisy sound into images to remove the noise has been proposed. We are attempting to remove train running noises and convert announcements into text in order to communicate train announcements to hearing disabilities. In the previous studies, the noise was removed by using U-Net with images converted from noisy sound. However, the quality of sound was not sufficient, since the restored sound was distorted. In this study, the optimal network model was built by adjusting conversion parameters of STFT and training parameters. The noise removal experiments from in-train announcements using data with multiple signal-to-noise ratios including low signal-to-noise ratios assuming in-train have been carried out. The recognition accuracy of noise-removed voice by voice recognition engine was improved, and the model with robustness could be built. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Spectrum / Noise Removal / U-Net / Train Running Noise / Robustness / Voice Recognition |
Paper # | SeMI2022-26 |
Date of Issue | 2022-07-06 (SeMI) |
Conference Information | |
Committee | NS / SR / RCS / SeMI / RCC |
---|---|
Conference Date | 2022/7/13(3days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | The Kanazawa Theatre + Online |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | Distributed Wireless Network, M2M (Machine-to-Machine),D2D (Device-to-Device),IoT(Internet of Things), etc |
Chair | Tetsuya Oishi(NTT) / Suguru Kameda(Hiroshima Univ.) / Kenichi Higuchi(Tokyo Univ. of Science) / Koji Yamamoto(Kyoto Univ.) / Shunichi Azuma(Nagoya Univ.) |
Vice Chair | Takumi Miyoshi(Shibaura Insti of Tech.) / Osamu Takyu(Shinshu Univ.) / Kentaro Ishidu(NICT) / Kazuto Yano(ATR) / Tomoya Tandai(Toshiba) / Fumihide Kojima(NICT) / Osamu Muta(Kyushu Univ.) / Kazuya Monden(Hitachi) / Yasunori Owada(NICT) / Shunsuke Saruwatari(Osaka Univ.) / Shunichi Azuma(Hokkaido Univ.) / Koji Ishii(Kagawa Univ.) |
Secretary | Takumi Miyoshi(NTT) / Osamu Takyu(Kogakuin Univ.) / Kentaro Ishidu(Mie Univ.) / Kazuto Yano(Tokai Univ.) / Tomoya Tandai(NTT) / Fumihide Kojima(Panasonic) / Osamu Muta(Univ. of Electro-Comm) / Kazuya Monden(Sharp) / Yasunori Owada(NTT DOCOMO) / Shunsuke Saruwatari(Tokyo Univ. of Agri. and Tech.) / Shunichi Azuma(Osaka Univ.) / Koji Ishii(CRIEPI) |
Assistant | Kotaro Mihara(NTT) / Taichi Ohtsuji(NEC) / WANG Xiaoyan(Ibaraki Univ.) / Akemi Tanaka(MathWorks) / Katsuya Suto(Univ. of Electro-Comm) / Manabu Sakai(Mitsubishi Electric) / Masashi Iwabuchi(NTT) / Tatsuki Okuyama(NTT DOCOMO) / Issei Kanno(KDDI Research) / Yuyuan Chang(Tokyo Inst. of Tech) / Yuki Matsuda(NAIST) / Akihito Taya(Aoyama Gakuin Univ.) / Takeshi Hirai(Osaka Univ.) / SHAN LIN(NICT) / Ryosuke Adachi(Yamaguchi Univ.) |
Paper Information | |
Registration To | Technical Committee on Network Systems / Technical Committee on Smart Radio / Technical Committee on Radio Communication Systems / Technical Committee on Sensor Network and Mobile Intelligence / Technical Committee on Reliable Communication and Control |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Investigation of noise removal using U-Net and voice recognition performance improvement |
Sub Title (in English) | for train running noise |
Keyword(1) | Spectrum |
Keyword(2) | Noise Removal |
Keyword(3) | U-Net |
Keyword(4) | Train Running Noise |
Keyword(5) | Robustness |
Keyword(6) | Voice Recognition |
1st Author's Name | Jian Lin |
1st Author's Affiliation | Kanagawa Institute of Technology(KAIT) |
2nd Author's Name | Shota Sano |
2nd Author's Affiliation | Kanagawa Institute of Technology(KAIT) |
3rd Author's Name | Yuusuke Kawakita |
3rd Author's Affiliation | Kanagawa Institute of Technology(KAIT) |
4th Author's Name | Tsuyoshi Miyazaki |
4th Author's Affiliation | Kanagawa Institute of Technology(KAIT) |
5th Author's Name | Hiroshi Tanaka |
5th Author's Affiliation | Kanagawa Institute of Technology(KAIT) |
Date | 2022-07-13 |
Paper # | SeMI2022-26 |
Volume (vol) | vol.122 |
Number (no) | SeMI-108 |
Page | pp.pp.34-39(SeMI), |
#Pages | 6 |
Date of Issue | 2022-07-06 (SeMI) |