| Presentation | 2002/10/23 Word Recognition Using the HMM Composition Method Which Suits a Signal-to-Noise Ratio Automatically Kenji TAKADA, Jun TOYAMA, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | HMM composition is known as one of the very effective technique for speech recognition under noisy environment. However, HMM composition cannot treat such a situation where Signal-to-Noise Ratio varies among frames. In this paper, we improve the original HMM composition technique so as to adapt such a variation. We do this by choosing appropriate values of the level adjustment coefficient on the basis of a maximum likelihood criterion applied to individual frames. Differences between the true values and the estimated values were measured and a word recognition experiment was conducted to demonstrate the effectivenss of this approach. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Word Recognition / Noisy Environment / HMM Composition / Level Adjustment Coefficient / Signal-to-Noise Ratio / Maximum Likelihood Procedure |
| Paper # | WIT2002-37 |
| Date of Issue |
| Conference Information | |
| Committee | WIT |
|---|---|
| Conference Date | 2002/10/23(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 | Well-being Information Technology(WIT) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Word Recognition Using the HMM Composition Method Which Suits a Signal-to-Noise Ratio Automatically |
| Sub Title (in English) | |
| Keyword(1) | Word Recognition |
| Keyword(2) | Noisy Environment |
| Keyword(3) | HMM Composition |
| Keyword(4) | Level Adjustment Coefficient |
| Keyword(5) | Signal-to-Noise Ratio |
| Keyword(6) | Maximum Likelihood Procedure |
| 1st Author's Name | Kenji TAKADA |
| 1st Author's Affiliation | Hokkaido University Graduate School of Engineering() |
| 2nd Author's Name | Jun TOYAMA |
| 2nd Author's Affiliation | Hokkaido University Graduate School of Engineering |
| Date | 2002/10/23 |
| Paper # | WIT2002-37 |
| Volume (vol) | vol.102 |
| Number (no) | 419 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |