Presentation 2003/4/17
Rubust Speech Recognition Based on HMM Composition and Noisy Frame Detection
Naoto ISAKA, Hiroshi OHSUGA, Takeshi YAMADA, Nobuhiko KITAWAKI, Futoshi ASANO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper proposes a new method for noise robust speech recognition, which is based on HMM composition and noisy frame detection. The proposed method detects noisy frames, and estimates their noise and SNR values by 2-steps detection. Recognition of input frames is performed by frame-wise model selection using the information in each frame. To evaluate the performance of the proposed method, connected-word recognition experiments were conducted. These results showed that the word correct rate of the proposed method is improved by a maximum of 19.5% compared to that of the use of multi-path models.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Robust Speech Recognition / HMM composition / Noisy Frame Detection / Non-speech Models
Paper # EA2003-1,SP2003-1
Date of Issue

Conference Information
Committee EA
Conference Date 2003/4/17(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 Engineering Acoustics (EA)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Rubust Speech Recognition Based on HMM Composition and Noisy Frame Detection
Sub Title (in English)
Keyword(1) Robust Speech Recognition
Keyword(2) HMM composition
Keyword(3) Noisy Frame Detection
Keyword(4) Non-speech Models
1st Author's Name Naoto ISAKA
1st Author's Affiliation Graduate School of Systems and Information Engineering, University of Tsukuba()
2nd Author's Name Hiroshi OHSUGA
2nd Author's Affiliation College of Information Sciences, University of Tsukuba
3rd Author's Name Takeshi YAMADA
3rd Author's Affiliation Institute of Information Sciences and Electronics, University of Tsukuba
4th Author's Name Nobuhiko KITAWAKI
4th Author's Affiliation Institute of Information Sciences and Electronics, University of Tsukuba
5th Author's Name Futoshi ASANO
5th Author's Affiliation National Institute of Advanced Industrial Science and Technology
Date 2003/4/17
Paper # EA2003-1,SP2003-1
Volume (vol) vol.103
Number (no) 24
Page pp.pp.-
#Pages 5
Date of Issue