Pronunciation modeling using a finite-state transducer representation
| Autor: | I. Lee Hetherington, Han Shu, Karen Livescu, Timothy J. Hazen |
|---|---|
| Rok vydání: | 2005 |
| Předmět: |
Linguistics and Language
Finite-state machine Computer science business.industry Communication Speech recognition Realization (linguistics) Word error rate Pronunciation computer.software_genre Speech processing Language and Linguistics Computer Science Applications Modeling and Simulation Component (UML) Expectation–maximization algorithm Computer Vision and Pattern Recognition Artificial intelligence Representation (mathematics) business computer Software Natural language processing |
| Zdroj: | Speech Communication. 46:189-203 |
| ISSN: | 0167-6393 |
| DOI: | 10.1016/j.specom.2005.03.004 |
| Popis: | The MIT summit speech recognition system models pronunciation using a phonemic baseform dictionary along with rewrite rules for modeling phonological variation and multi-word reductions. Each pronunciation component is encoded within a finite-state transducer (FST) representation whose transition weights can be trained using an EM algorithm for finite-state networks. This paper explains the modeling approach we use and the details of its realization. We demonstrate the benefits and weaknesses of the approach both conceptually and empirically using the recognizer for our jupiter weather information system. Our experiments demonstrate that the use of phonological rewrite rules within our system achieves word error rate reductions between 4% and 9% over different test sets when compared against a system using no phonological rewrite rules. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect