Presentation | 2000/12/14 Unsupervised training based on the sufficient HMM statistics from selected speakers Shinichi Yoshizawa, Akira Baba, Kanako Matsunami, Yuichiro Mera, Miichi Yamada, Kiyohiro Shikano, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | This paper describes an efficient method of unsupervised training. This method is based on (1) selecting a subset of speakers who are acoustically close to a test speaker, and (2) calculating adapted model parameters according to the previously stored sufficient HMM statistics of the selected speakers' data. In this method, only a few unsupervised test speaker's data are required. Also, by using the sufficient HMM statistics of the selected speakers' data, a quick training can be done. Compared with a pre-clustering method, the proposed method can obtain a more optimal speaker cluster because the clustering result is determined according to test speaker's data on-line. Experiment results show that the proposed method attains better improvement than MLLR from the speaker-independent model. Moreover the proposed method utilizes only one unsupervised sentence utterance, while MLLR usually utilizes more than ten supervised sentence utterances. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | acoustic model / speaker adaptation / sufficient statistics / unsupervised |
Paper # | NLC2000-41,SP2000-89 |
Date of Issue |
Conference Information | |
Committee | NLC |
---|---|
Conference Date | 2000/12/14(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) | Unsupervised training based on the sufficient HMM statistics from selected speakers |
Sub Title (in English) | |
Keyword(1) | acoustic model |
Keyword(2) | speaker adaptation |
Keyword(3) | sufficient statistics |
Keyword(4) | unsupervised |
1st Author's Name | Shinichi Yoshizawa |
1st Author's Affiliation | Laboratories of Image Information Science and Technology() |
2nd Author's Name | Akira Baba |
2nd Author's Affiliation | Laboratories of Image Information Science and Technology |
3rd Author's Name | Kanako Matsunami |
3rd Author's Affiliation | Nara Institute of Science and Technology |
4th Author's Name | Yuichiro Mera |
4th Author's Affiliation | Nara Institute of Science and Technology |
5th Author's Name | Miichi Yamada |
5th Author's Affiliation | Nara Institute of Science and Technology |
6th Author's Name | Kiyohiro Shikano |
6th Author's Affiliation | Nara Institute of Science and Technology |
Date | 2000/12/14 |
Paper # | NLC2000-41,SP2000-89 |
Volume (vol) | vol.100 |
Number (no) | 520 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |