The Phoneme Automatic Segmentation Algorithms Study of Tibetan Lhasa Words Continuous Speech Stream

Abstract:

In this paper, we adopt two methods to voice phoneme segmentation when building Tibetan corpus: One is the traditional artificial segmentation method, one is the automatic segmentation method based on the Mono prime HMM model. And experiments are performed to analyze the accuracy of both methods of segmentations. The results showed: Automatic segmentation method based tone prime HMM model helps to shorten the cycle of building Tibetan corpus, especially in building a large corpus segmentation and labeling a lot of time and manpower cost savings, and have greatly improved the accuracy and consistency of speech corpus annotation information.

• Periodical: Advanced Materials Research (Volumes 765-767)
• Pages: 2051-2054
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.765-767.2051
• Online since: September 2013
• Authors: Jin Xi Zhang, Hong Zhi Yu, Ning Ma, Zhao Yao Li
• Keywords: Corpus, Phoneme Automatic Segmentation, Tibetan

References

[1] WANG Li-juan, CAO Zhi-gang. Automatic Phonetic Segmentation Using HMM Model [J]. Journal of Data Acquisition & Processing, 2005, 20(4): 381-384.

[2] CHEN Kai, CAI Pei-qi. Silence Insertion in HMM-based Chinese Automatic Segmentation [J]. Computer Engineering. 2004, 30(9): 40-41.

[3] ASKAR Rozi. Research and Implementation of HMM based Uyghur Speech Synthesis System [D]. Xinjiang University. (2008).

[4] Gao Lu, Yu Hongzhi, et al. Study on SAMPA_ST for Lhasa Tibetan and Realization of Automatic Labeling System[C]. IASP 2010. Vol I, pp.133-137.

[5] WANG Li-juan, CAO Zhi-gang. Automatic Segmentation for TTS Units [J]. Microelectronics and Computer, 2005, 22(12): 8-11.

[6] Htkbookhttp: /users. ece. gatech. edu/~antonio/htkbook/htkbook. html[OL].