Generation and selection of English pronunciation variations using knowledge-based rules and speech recognition techniques
0
조회
0
다운로드
1
인용
0
참고문헌
초록
Abstract
This paper describes a method of designing a pronunciation dictionary which contains an optimal number of pronunciation variants for each word entry. The two key processes, generation and selection of variants, are designed to be performed automatically so that the final version of pronunciation dictionary can be practically used in speech recognition systems. For generation of variants both prescriptive and descriptive rule extraction methods are employed. First, various optional phonological processes are extracted from literature. Second, phone strings obtained by hand labelling are compared with base-form pronunciation strings and the systematic differences are set as rules. After generation of as many variants as possible, it is determined whether each variant deserves to be kept in the dictionary through relative frequency measure and heuristic decision-making criteria. Through experiments, the variant dictionary is found to play a meaningful role in enhancing performance of automatic speech recognition.
참고문헌 (19)
- BATES, "Modelling pronunciation variation in conversational speech using prosody,Proceedings of the Workshop on Prosody in Speech Recognition and Understanding", pp.17~22 (2001)
- BINNENPOORTE, "Multiword expressions in spoken language: an exploratory study on pronunciation variation", 19, pp.433~449 (2005)
- BONAVENTURA, "Speech recognition methods for non-native pronunciation variations", pp.14~22 (1998)
- BYRNE, "Pronunciation Modelling for Conversational Speech Recognition: A Status Report from WS97, Proceedings of 1997 IEEE Workshop on Automatic Speech Recognition and Understanding" (1997), Santa Barbara, California, U.S.A
- GORONZY, "Generating non-native pronunciation variants for lexicon adaptation", 42, pp.109~123 (2004)
- GREENBERG, "The Switchboard Transcription Project,in Research Report #24, Large Vocabulary Continuous Speech Recognition Summer Research Workshop Technical Report Series,Center for Language and Speech Processing" (1997)
- Strik, "Speaking in shorthand--a syllable-centric perspective for understanding pronunciation variation", pp.47~56 (1998)
- HEINE, "An HMM-based probabilistic lexicon", pp.57~62 (1998)
- MICHAEL, "Phonology in generative grammar" (1994)
- KESSENS, "A data-driven method for modeling pronunciation variation", 40(4), pp.517~534 (2003)
- KREIDLER, "The pronunciation of English a course book in phonology" (1999), Blackwell
- ROACH, "English Phonetics and phonology: a practical course" (2004), Cambridge University Press
- SILVERMAN, "A critical introduction to phonology of sound, mind, and body" (2006), Continuum
- STRIK, "Modeling pronunciation variation for automatic speech recognition, Rolduc, The Nethelands. European Speech Communication Association" (1998)
- TATHAM, "Speech Production and perception" (2006), Palgrave Macmillan
- WARD, "Lexicon adaptation for LVCSR: speaker idiosyncracies, non-native speakers, and pronunciation choice,ISCA Workshop on Pronunciation Modeling and Lexicon Adaptation" (2002), Estes Park, Colorado
- WESTER, "Pronunciation modeling for ASR-knowledge-based and data-derived methods", 17(1), pp.69~85 (2003)
- YANG, "On the importance of exception and cross-word rules for the data-driven creation of Lexica for ASR. Proceedings of 11th ProRisc Workshop,Veldhoven, The Netherlands", pp.589~593 (2000)
- YOUNG, STEVE, "HTK Book,Entropic" (1996)