Abstract
Chinese character is composed of spatial arrangement of strokes. A portion of these strokes combines to form phonetic component, which provides a clue to the pronunciation of the entire character, the others combine to form semantic component, which indicates semantic level information for speech context. How closely the connection between the internal strokes of Chinese characters and speech? In this paper, we propose Speech2Stroke, a end-to-end model that exploits the phonetic and morphologic level information of pictographic words. Specifically, we generate strokes directly from the speech by Speech2Stroke. The performance of Speech2Stroke is evaluated by the specific stroke error rate(SER). The SER of the optimal model can achieve 20.61%. Through the experiments and analysis, we show that our model has the ability to capture the alignment between audio and the internal structures of pictographic characters.
This work was supported by NSFC Grant No. 61832008, 61772413, 61802299, and 61672424.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Standardization of stroke order for modern Chinese homepage (2005). http://www.moe.gov.cn/s78/A19/yxs_left/moe_810/s230/201001/t20100115_75615.html, Accessed 2 May 2020
Amodei, D., et al.: Deep speech 2: end-to-end speech recognition in English and Mandarin. In: International Conference on Machine Learning, pp. 173–182 (2016)
Bu, H., Du, J., Na, X., Wu, B., Zheng, H.: Aishell-1: an open-source mandarin speech corpus and a speech recognition baseline. In: 2017 20th Conference of the Oriental Chapter of the International Coordinating Committee on Speech Databases and Speech I/O Systems and Assessment (O-COCOSDA), pp. 1–5. IEEE (2017)
Cao, S., Lu, W., Zhou, J., Li, X.: cw2vec: learning Chinese word embeddings with stroke n-gram information. In: Thirty-Second AAAI Conference on Artificial Intelligence (2018)
Caruana, R., Lawrence, S., Giles, C.L.: Overfitting in neural nets: backpropagation, conjugate gradient, and early stopping. In: Advances in Neural Information Processing Systems, pp. 402–408 (2001)
Cho, K., et al.: Learning phrase representations using RNN encoder-decoder for statistical machine translation. arXiv preprint arXiv:1406.1078 (2014)
Graves, A., Fernández, S., Gomez, F., Schmidhuber, J.: Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks. In: Proceedings of the 23rd International Conference on Machine Learning, pp. 369–376 (2006)
Hannun, A.Y., Maas, A.L., Jurafsky, D., Ng, A.Y.: First-pass large vocabulary continuous speech recognition using bi-directional recurrent DNNs. arXiv preprint arXiv:1408.2873 (2014)
Hochreiter, S., Schmidhuber, J.: Long short-term memory. Neural Comput. 9(8), 1735–1780 (1997)
Ioffe, S., Szegedy, C.: Batch normalization: Accelerating deep network training by reducing internal covariate shift. arXiv preprint arXiv:1502.03167 (2015)
Levenshtein, V.I.: Binary codes capable of correcting deletions, insertions, and reversals. Soviet Phys. Doklady 10, 707–710 (1966)
Meng, Y., Wet al.: Glyce: glyph-vectors for Chinese character representations. In: Advances in Neural Information Processing Systems, pp. 2742–2753 (2019)
Chinese. BLS. Macmillan Education UK, London (1999). https://doi.org/10.1007/978-1-349-27306-5_9
Paszke, A., et al.: Pytorch: an imperative style, high-performance deep learning library. In: Advances in Neural Information Processing Systems, pp. 8024–8035 (2019)
Sainath, T.N., Mohamed, A.R., Kingsbury, B., Ramabhadran, B.: Deep convolutional neural networks for LVCSR. In: 2013 IEEE International Conference on Acoustics, Speech and Signal Processing, pp. 8614–8618. IEEE (2013)
Su, T.R., Lee, H.Y.: Learning Chinese word representations from glyphs of characters. In: Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 264–273 (2017)
Sutskever, I., Martens, J., Dahl, G., Hinton, G.: On the importance of initialization and momentum in deep learning. In: International Conference on Machine Learning, pp. 1139–1147 (2013)
Werbos, P.J.: Backpropagation through time: what it does and how to do it. Proc. IEEE 78(10), 1550–1560 (1990)
Yin, R., Wang, Q., Li, P., Li, R., Wang, B.: Multi-granularity Chinese word embedding. In: Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pp. 981–986 (2016)
Yu, J., Jian, X., Xin, H., Song, Y.: Joint embeddings of Chinese words, characters, and fine-grained subcharacter components. In: Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 286–291 (2017)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering
About this paper
Cite this paper
Zhang, Y. et al. (2021). Speech2Stroke: Generate Chinese Character Strokes Directly from Speech. In: Gao, H., Wang, X., Iqbal, M., Yin, Y., Yin, J., Gu, N. (eds) Collaborative Computing: Networking, Applications and Worksharing. CollaborateCom 2020. Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering, vol 349. Springer, Cham. https://doi.org/10.1007/978-3-030-67537-0_6
Download citation
DOI: https://doi.org/10.1007/978-3-030-67537-0_6
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-67536-3
Online ISBN: 978-3-030-67537-0
eBook Packages: Computer ScienceComputer Science (R0)