| Presentation | 2007/12/13 Automatic Speech Recognition of Congressional Speech Based on Statistical Style Transformation of Language Model and Pronunciation Model Yuya AKITA, Tatsuya KAWAHARA, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | For automatic speech recognition (ASR) of spontaneous speech such as congressional meetings, we have been proposing statistical transformation methods of language model and pronunciation model. In these methods, differences between faithful transcripts and orthographical transcripts are statistically extracted. Then, transformation models which consist of probabilistic transformation patterns are derived from the statistics for language model and pronunciation model. For language model, the transformation model predicts spoken-style N-gram entries with estimated occurrence counts. For pronunciation model, pronunciation variants and their probabilities are predicted by the transformation model. The language model and pronunciation model generated by the proposed methods were evaluated on ASR of committee meetings of Japanese National Congress (Diet),and realized absolute reduction of word error rates by 0.6-0.7% and 1.0%, respectively, compared with models produced by conventional methods. Finally, total reduction of 1.7% was obtained by combining both models. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Spontaneous speech / Speech recognition / Language model / Lexicon / Speaking style |
| Paper # | NLC2007-43,SP2007-106 |
| Date of Issue |
| Conference Information | |
| Committee | NLC |
|---|---|
| Conference Date | 2007/12/13(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) | Automatic Speech Recognition of Congressional Speech Based on Statistical Style Transformation of Language Model and Pronunciation Model |
| Sub Title (in English) | |
| Keyword(1) | Spontaneous speech |
| Keyword(2) | Speech recognition |
| Keyword(3) | Language model |
| Keyword(4) | Lexicon |
| Keyword(5) | Speaking style |
| 1st Author's Name | Yuya AKITA |
| 1st Author's Affiliation | Academic Center for Computing and Media Studies, Kyoto University() |
| 2nd Author's Name | Tatsuya KAWAHARA |
| 2nd Author's Affiliation | Academic Center for Computing and Media Studies, Kyoto University |
| Date | 2007/12/13 |
| Paper # | NLC2007-43,SP2007-106 |
| Volume (vol) | vol.107 |
| Number (no) | 405 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |