Using Stress Information in Large Vocabulary Speech Recognition
Abstract
Using stress information in a Markov source-based large vocabulary speech recognition system provides a way to examine a nonlocal cue which, is generally poorly represented by the Markov source model. In this paper, we present an algorithm for estimating the stress pattern based on syllable durations and short-time energies. The output also gives the probability of the correctness of the estimated stress pattern. The parameters are first normalized in an attempt to reduce. variability due to different linguistic contexts. The stress pattern is then estimated based on a statistical approach. After initial training, tests on a new word list yielded 95% correct detection of the syllable carrying the primary stress. Finally. inclusion of this algorithm in a large vocabulary isolated word recognition system contributes to its accuracy.
Downloads
Published
How to Cite
Issue
Section
License
Copyright on articles is held by the author(s). The corresponding author has the right to grant on behalf of all authors and does grant on behalf of all authors, a worldwide exclusive licence (or non-exclusive license for government employees) to the Publishers and its licensees in perpetuity, in all forms, formats and media (whether known now or created in the future)
i) to publish, reproduce, distribute, display and store the Contribution;
ii) to translate the Contribution into other languages, create adaptations, reprints, include within collections and create summaries, extracts and/or, abstracts of the Contribution;
iii) to exploit all subsidiary rights in the Contribution,
iv) to provide the inclusion of electronic links from the Contribution to third party material where-ever it may be located;
v) to licence any third party to do any or all of the above.
