Synopses & Reviews
This original volume describes the Spoken Language Translator (SLT), one of the first major automatic speech translation projects. The SLT system can translate between English, French, and Swedish in the domain of air travel planning, using a vocabulary of about 1500 words, and with an accuracy of about 75%. The authors detail the language processing components, largely built on top of the SRI Core Language Engine, using a combination of general grammars and techniques that allow them to be rapidly customized to specific domains. They base speech recognition on Hidden Markov Mode technology, and use versions of the SRI DECIPHER system. This account of SLT is an essential resource for researchers interested in knowing what is achievable in spoken-language translation today.
Review
"...this book is a useful resource for those who are interested in knowing how to build a speech-to-speech translation system, especially for those who are interested in grammar-based language processing approaches." Computational Linguistics
Table of Contents
1. Introduction; Part I. Language Processing and Corpora: 2. Translation using the core language engine; 3. Grammar specialization; 4. Choosing between interpretations; 5. The TreeBanker; 6. Acquisition of lexical entries; 7. Spelling and morphology; 8. Corpora and data collection; Part II. Linguistic Coverage: 9. English coverage; 10. French coverage; 11. Swedish coverage; 12. Transfer coverage; 13. Rational re-use of linguistic data; Part III. Speech Processing: 14. Speech recognition; 15. Acoustic modeling; 16. Language modeling for multilingual speech translation; 17. Porting a recognizer to a new language; 18. Multiple dialects and languages; 19. Common speech-language issues; Part IV. Evaluation and Conclusions: 20. Evaluation; 21. Conclusions; Appendix A. The mathematics of discriminant scores; Appendix B. Notation for QLF-based processing.