Crandem systems: Conditional random field acoustic models for hidden Markov models
Fosler, E.L.
Morris, J.
Dept. of Comput. Sci. & Eng., Ohio State Univ., Columbus, OH;
This paper appears in: Acoustics, Speech and Signal Processing, 2008. ICASSP 2008. IEEE International Conference on
Publication Date: March 31 2008-April 4 2008
On page(s): 4049-4052
Location: Las Vegas, NV,
ISSN: 1520-6149
ISBN: 978-1-4244-1483-3
INSPEC Accession Number: 9973885
Digital Object Identifier: 10.1109/ICASSP.2008.4518543
Current Version Published: 2008-05-12
Abstract
In recent years, Conditional Random Fields (CRFs) have been examined as a statistical model for speech recognition. In this paper, we explore the use of features derived via CRFs as inputs to a Tandem- style HMM ASR system (that is, a Crandem system). We present a model for deriving frame-level posterior features via CRFs to use in Crandem modeling and additionally provide experimental results that show the Crandem system can slightly significantly outperform both a comparable Tandem system and a comparable CRF system on the task of phone recognition.
Index
Terms
Available to subscribers and IEEE members.
References
Available to subscribers and IEEE members.
Citing Documents
Available to subscribers and IEEE members.
Cart
