| Presentation | 2005/7/15 Reverberation modeling on power spectral trajectory for distant speech recognition Tasuku TAKEI, Hiroshi MATSUMOTO, Kazumasa YAMAMOTO, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In order to reduce the influence of reverberation in distance speech recogintion, this paper examines a reverberation model on the power trajectory domain at the output of a mel-filter in the MFCC analysis. The model parameters consist of the decay rate representing reverberation, the ratio of reverberant power to the direct sound, and the frequency response of the channel including some parts of coloration. These model parameters are estimated for each frequency band based on a minimum mean square error of log-power trajectory using pairs of clean speech and their reverberant counterparts. HMMs trained by MFCC derived from synthesized power trajectory with the estimated parameters attained a few percet lower recognition accuracy than that obtaind by actual reverberant HMMs. Furthermore, the dereververation by an inverse filter based on the model and post- processing by flooring and smoothing improved the recognition accuracy by about 10% in Acc. compared to non-processed speech. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Reverberation / Dereverberation / Distant speech recognition / Power trajectory / Hands-free speech recognition |
| Paper # | SP2005-43 |
| Date of Issue |
| Conference Information | |
| Committee | SP |
|---|---|
| Conference Date | 2005/7/15(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 | Speech (SP) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Reverberation modeling on power spectral trajectory for distant speech recognition |
| Sub Title (in English) | |
| Keyword(1) | Reverberation |
| Keyword(2) | Dereverberation |
| Keyword(3) | Distant speech recognition |
| Keyword(4) | Power trajectory |
| Keyword(5) | Hands-free speech recognition |
| 1st Author's Name | Tasuku TAKEI |
| 1st Author's Affiliation | Faculuty of Engineering, Shinshu University() |
| 2nd Author's Name | Hiroshi MATSUMOTO |
| 2nd Author's Affiliation | Faculuty of Engineering, Shinshu University |
| 3rd Author's Name | Kazumasa YAMAMOTO |
| 3rd Author's Affiliation | Faculuty of Engineering, Shinshu University |
| Date | 2005/7/15 |
| Paper # | SP2005-43 |
| Volume (vol) | vol.105 |
| Number (no) | 199 |
| Page | pp.pp.- |
| #Pages | 5 |
| Date of Issue | |