Presentation 2009-12-17
Speaker Dependent Speech Recognition from Marker Training Data
Shun MATSUI, Nozomu HAMADA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) As a visual audio fusing scheme for speech recognition in noisy environment, this paper aims to establish appropriate visual-side feature using speech recognition for a specific speaker. Three visual feature vectors are used in the learning process, such as the positions of markers attached on face, the mouth contour shape, and eigen lip images of the speaker's utterance. In test process, the latter two features are used for recognition based on the obtained association with maker locations in the learning. The proposed method gives higher recognition rate.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) word recognition / lip-reading / visual feature / marker / lip contour / eigenlip
Paper # SIS2009-40
Date of Issue

Conference Information
Committee SIS
Conference Date 2009/12/10(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 Smart Info-Media Systems (SIS)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Speaker Dependent Speech Recognition from Marker Training Data
Sub Title (in English)
Keyword(1) word recognition
Keyword(2) lip-reading
Keyword(3) visual feature
Keyword(4) marker
Keyword(5) lip contour
Keyword(6) eigenlip
1st Author's Name Shun MATSUI
1st Author's Affiliation School of Integrated Design Engineering, Keio University()
2nd Author's Name Nozomu HAMADA
2nd Author's Affiliation School of Integrated Design Engineering, Keio University
Date 2009-12-17
Paper # SIS2009-40
Volume (vol) vol.109
Number (no) 338
Page pp.pp.-
#Pages 6
Date of Issue