Korean Phoneme Boundary Detection Based on Time Domain Metrics
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This paper proposes a novel Korean phoneme boundary detection method that can be applied to phoneme-basedKorean speech recognition systems. The proposed method employs two time-domain metrics—volatility andbulk metrics—as the foundation for phoneme boundary detection. The input speech signal is divided into blocksof 300 integer samples. For each block, the volatility metric is computed that adds up all the changes betweenneighboring samples within the block. A bulk is a grouping of consecutive samples with the same sign. Foreach bulk, two bulk metrics are calculated: bulk size and bulk length. Three dedicated algorithms that utilizeboth types of metrics are used to detect phoneme boundaries by recognizing vowels, voiced consonants, andvoiceless consonants in turn. The experimental results show that the proposed method can significantly reducethe error rate compared to an existing boundary detection method.
Abstract
This paper proposes a novel Korean phoneme boundary detection method that can be applied to phoneme-basedKorean speech recognition systems. The proposed method employs two time-domain metrics—volatility andbulk metrics—as the foundation for phoneme boundary detection. The input speech signal is divided into blocksof 300 integer samples. For each block, the volatility metric is computed that adds up all the changes betweenneighboring samples within the block. A bulk is a grouping of consecutive samples with the same sign. Foreach bulk, two bulk metrics are calculated: bulk size and bulk length. Three dedicated algorithms that utilizeboth types of metrics are used to detect phoneme boundaries by recognizing vowels, voiced consonants, andvoiceless consonants in turn. The experimental results show that the proposed method can significantly reducethe error rate compared to an existing boundary detection method.
참고문헌 (13)
- S. K. Gaikwad, "A review on speech recognition technique", International Journal of Computer Applications, 10(3), pp.16~24 (2010)[DOI]
- Y. Y. Seo, "Phoneme segmentation in consideration of speech feature in Korean speech recognition", Journal of Internet Computing and Services, 2(1), pp.31~38 (2001)
- M. Y. Nam, "Recognition of Korean fricatives and affricates using modified Teager energy measurement method", Proceedings of the IEEK Conference, 15(1), pp.23~26 (1993)
- S. Brognaux, "HMM-based speech segmentation : improvements of fully automatic approaches", IEEE, 2016, pp.24(1), 5~15
- P. B. Ramteke, "Phoneme boundary detection from speech : a rule based approach", Speech Communication, 107, pp.1~17 (2019)[DOI]
- F. Kreuk, "Phoneme boundary detection using learnable segmental features", Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp.8089~8093 (2020)
- K. K. Ravi, "Phoneme segmentation-based unsupervised pattern discovery and clustering of speech signals", Circuits, Systems, and Signal Processing, 41(4), pp.2088~2117 (2022)[DOI]
- B. Lin, "Learning acoustic frame labeling for phoneme segmentation with regularized attention mechanism", Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp.7882~7886 (2022)
- H. Frihia, "HMM", SVM segmentation and labelling of Arabic speech for speech recognition applications, 2017, pp.20(3), 563~573
- A. E. Sakran, "A review : automatic speech segmentation", International Journal of Computer Science and Mobile Computing, 6(4), pp.308~315 (2017)
- Y. Lee, "Phoneme segmentation using phoneme combination and formant scaling in Korean", M. S., Department of Computer Engineering, pp.2003
- N. Lachachi, "Unsupervised phoneme segmentation based on main energy change for Arabic speech", Journal of Telecommunications and Information Technology, 2017(1), pp.12~20 (2017)[DOI]
- J. W. Lee, "Phoneme segmentation based on volatility and bulk indicators in Korean speech recognition", KIISE Transactions on Computing Practices, 21(10), pp.631~638 (2015)[DOI]