Modeling continuous visual speech using boosted viseme models
In this paper, a novel connected-viseme approach for modeling continuous visual speech is presented. The approach adopts AdaBoost-HMMs as the viseme models. Continuous visual speech is modeled by connecting the viseme models using level building algorithm. The approach is applied to identify words...
Saved in:
Main Authors: | Dong, Liang, Foo, Say Wei, Yong, Lian |
---|---|
Other Authors: | School of Electrical and Electronic Engineering |
Format: | Conference or Workshop Item |
Language: | English |
Published: |
2009
|
Subjects: | |
Online Access: | https://hdl.handle.net/10356/91029 http://hdl.handle.net/10220/6002 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Language: | English |
Similar Items
-
Cross-speaker viseme mapping using hidden Markov models
by: Dong, Liang, et al.
Published: (2009) -
A boosted multi-HMM classifier for recognition of visual speech elements
by: Foo, Say Wei, et al.
Published: (2009) -
A two-channel training algorithm for hidden Markov model to identify visual speech elements
by: Foo, Say Wei, et al.
Published: (2009) -
A simplified viterbi matching algorithm for word partition in visual speech processing
by: Foo, Say Wei, et al.
Published: (2009) -
Recognition of visual speech elements using adaptively boosted hidden Markov models
by: Foo, Say Wei, et al.
Published: (2009)