비교사 분할 및 병합으로 구한 의사형태소 음성인식 단위의 성능
Performance of Pseudomorpheme-Based Speech Recognition Units Obtained by Unsupervised Segmentation and Merging
1
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
This paper proposes a new method to determine the recognition units for large vocabulary continuous speech recognition(LVCSR) in Korean by applying unsupervised segmentation and merging. In the proposed method, a text sentence issegmented into morphemes and position information is added to morphemes. Then submorpheme units are obtained bysplitting the morpheme units through the maximization of posterior probability terms. The posterior probability terms arecomputed from the morpheme frequency distribution, the morpheme length distribution, and the morphemefrequency-of-frequency distribution. Finally, the recognition units are obtained by sequentially merging the submorpheme pairwith the highest frequency. Computer experiments are conducted using a Korean LVCSR with a 100k word vocabulary and atrigram language model obtained by a 300 million eojeol (word phrase) corpus. The proposed method is shown to reduce theout-of-vocabulary rate to 1.8% and reduce the syllable error rate relatively by 14.0%.
Abstract
This paper proposes a new method to determine the recognition units for large vocabulary continuous speech recognition(LVCSR) in Korean by applying unsupervised segmentation and merging. In the proposed method, a text sentence issegmented into morphemes and position information is added to morphemes. Then submorpheme units are obtained bysplitting the morpheme units through the maximization of posterior probability terms. The posterior probability terms arecomputed from the morpheme frequency distribution, the morpheme length distribution, and the morphemefrequency-of-frequency distribution. Finally, the recognition units are obtained by sequentially merging the submorpheme pairwith the highest frequency. Computer experiments are conducted using a Korean LVCSR with a 100k word vocabulary and atrigram language model obtained by a 300 million eojeol (word phrase) corpus. The proposed method is shown to reduce theout-of-vocabulary rate to 1.8% and reduce the syllable error rate relatively by 14.0%.
참고문헌 (21)
- Kwon, O. -W., "Korean large vocabulary continuous speech recognition using pseudomorpheme units", Proc. EUROSPEECH, pp.483~486 (1999)
- Yu, H. -J., "Automatic recognition of Korean broadcast news speech", Proc. ICSLP (1998)
- Kwon, O. -W., "Korean large vocabulary continuous speech recognition with morpheme-based recognition units", Speech Communication, 39(3), pp.287~300 (2003)
- Creutz, M., "Unsupervised discovery of Morphemes", Proc. ACL-02 Workshop on Morphological and Phonological Learning, pp.21~30 (2002)
- 김영택, "자연언어처리" (2001), 생능출판사
- Creutz, M., "Induction of the Morphology of Natural Language: Unsupervised Morpheme Segmentation with Application to Automatic Speech Recognition", Ph.D., pp.2006
- Schuster, M., "Japanese and Korean voice search", Proc. ICASSP, pp.5149~5152 (2012)
- Creutz, M., "Morfessor in the Morpho Challenge", Proc. PASCAL Challenge Workshop on Unsupervised Segmentation of Words into Morphemes (2006)
- Creutz, M., "Unsupervised segmentation of words using prior distributions of morph length and frequency", Proc. ACL-03, pp.280~287 (2003)
- Siivola, V., "Unlimited vocabulary speech recognition based on morphs discovered in an unsupervised manner", Proc. EUROSPEECH, pp.2293~2296 (2003)
- Hirsimaki, T., "Unlimited vocabulary speech recognition with morph language models applied to Finnish", Computer Speech & Language, 20(4), pp.515~541 (2006)
- Kwon, O. -W., "Development of a Korean large vocabulary continuous speech recognition platform (ECHOS)", Proc. O-COCOSDA, pp.108~111 (2007)
- Stolcke, A., "SRILM - An extensible language modeling toolkit", Proc. INTERSPEECH, pp.901~904 (2002)
- Povey, D., "The Kaldi speech recognition toolkit", Proc. ASRU, pp.1~4 (2011)
- "Downloading Kaldi"[URL]
- 박종렬, "한국어 음성 인식을 위한 음성 데이터 수집", 음향학회지, 14(4), pp.74~81 (1995)
- 최인정, "대용량 한국어 연속음성인식 시스템 개발", 음향학회지, 14(5), pp.44~50 (1995)
- Jurafsky, D., "Speech and Language Processing", (2e), pp.95~ (2008)
- "Google Voice Recognition" (2012)[URL]
- "Zipf's law"[URL]
- Jurafsky, D., "Speech and Language Processing, 2e. 4.5.2 Good-Turing Discounting", ~ (2008)