| Presentation | 2010-12-20 Noise suppression method based on bias-residual component decomposition and optimization Masakiyo FUJIMOTO, Shinji WATANABE, Tomohiro NAKATANI, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In this paper, we propose a non-stationary noise estimation method based on bias-residual component decomposition, and show that the proposed method improved speech recognition accuracy compared with the noise estimation methods without component decomposition. The proposed decomposition method clarifies the role of each component, thus enabling us to apply a suitable parameter estimation technique to each component. In this paper, the bias component is estimated by the EM algorithm with the entire observed signal sequence. On the other hand, the residual component is sequentially estimated by multiplying the extended Kalman filter with the auto-regressive process. In addition, the proposed noise estimation. As further improvement, we also evaluate the speech recognition by using a minimum-time frame word error. It minimizes word error rate by combining various hypotheses given by multiple parameter settings of the proposed noise estimation. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | speech recognition / noise suppression / non-stationary noise / component decomposition |
| Paper # | NLC2010-18,SP2010-91 |
| Date of Issue |
| Conference Information | |
| Committee | NLC |
|---|---|
| Conference Date | 2010/12/13(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Natural Language Understanding and Models of Communication (NLC) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Noise suppression method based on bias-residual component decomposition and optimization |
| Sub Title (in English) | |
| Keyword(1) | speech recognition |
| Keyword(2) | noise suppression |
| Keyword(3) | non-stationary noise |
| Keyword(4) | component decomposition |
| 1st Author's Name | Masakiyo FUJIMOTO |
| 1st Author's Affiliation | NTT Communication Science Laboratories, NTT Corporation() |
| 2nd Author's Name | Shinji WATANABE |
| 2nd Author's Affiliation | NTT Communication Science Laboratories, NTT Corporation |
| 3rd Author's Name | Tomohiro NAKATANI |
| 3rd Author's Affiliation | NTT Communication Science Laboratories, NTT Corporation |
| Date | 2010-12-20 |
| Paper # | NLC2010-18,SP2010-91 |
| Volume (vol) | vol.110 |
| Number (no) | 356 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |