Abstract
In this work we present simple grapheme-based system for low-resource speech recognition using Babel data for Turkish spontaneous speech (80 h). We have investigated different neural network architectures performance, including fully-convolutional, recurrent and ResNet with GRU. Different features and normalization techniques are compared as well. We also proposed CTC-loss modification using segmentation during training, which leads to improvement while decoding with small beam size.
Our best model achieved word error rate of 45.8%, which is the best reported result for end-to-end systems using in-domain data for this task, according to our knowledge.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
CTC Decoder for PyTorch. https://github.com/parlance/ctcdecode
Kaldi Recipe Results for Turkish Language. https://github.com/kaldi-asr/kaldi/blob/master/egs/babel/s5d/results/results.105-turkish-fullLP.official.conf.jtrmal1%40jhu.edu.2015-11-28T144317-0500
Sclite Scoring Package. http://www1.icsi.berkeley.edu/Speech/docs/sctk-1.2/sclite.htm
The SRI Language Modeling Toolkit. http://www.speech.sri.com/projects/srilm/
Turkish Alphabet. https://en.wikipedia.org/wiki/Turkish_alphabet
Alumäe, T., et al.: The 2016 BBN Georgian telephone speech keyword spotting system. In: Proceedings of ICASSP, pp. 5755–5759 (2017)
Amodei, D., et al.: Deep speech 2: end-to-end speech recognition in English and Mandarin (2015). arxiv:1512.02595
Chan, W., Jaitly, N., Le, Q.V., Vinyals, O.: Listen, attend and spell: a neural network for large vocabulary conversational speech recognition. In: Proceedings of ICASSP, pp. 4960–4964 (2016)
Chernykh, G., Korenevsky, M., Levin, K., Ponomareva, I., Tomashenko, N.: State level control for acoustic model training. In: Ronzhin, A., Potapova, R., Delic, V. (eds.) SPECOM 2014. LNCS (LNAI), vol. 8773, pp. 435–442. Springer, Cham (2014). https://doi.org/10.1007/978-3-319-11581-8_54
Collobert, R., Puhrsch, C., Synnaeve, G.: Wav2letter: an end-to-end ConvNet-based speech recognition system (2016). arxiv:1609.03193
Dalmia, S., Sanabria, R., Metze, F., Black, A.W.: Sequence-based multi-lingual low resource speech recognition (2018). arxiv:1802.07420
Gales, M.J.F., Knill, K.M., Ragni, A.: Low-resource speech recognition and keyword-spotting. In: Karpov, A., Potapova, R., Mporas, I. (eds.) SPECOM 2017. LNCS (LNAI), vol. 10458, pp. 3–19. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-66429-3_1
Graves, A.: Sequence transduction with recurrent neural networks (2012). arxiv:1211.3711
Graves, A., Fernández, S., Gomez, F., Schmidhuber, J.: Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks. In: Proceedings of ICML, pp. 369–376 (2006)
Hannun, A.Y., et al.: Deep speech: scaling up end-to-end speech recognition (2014). arxiv:1412.5567
Khokhlov, Y.Y., et al.: The STC keyword search system for OpenKWS 2016 evaluation. In: Proceedings of INTERSPEECH, pp. 3602–3606 (2017)
Khomitsevich, O., Mendelev, V., Tomashenko, N., Rybin, S., Medennikov, I., Kudubayeva, S.: A bilingual Kazakh-Russian system for automatic speech recognition and synthesis. In: Ronzhin, A., Potapova, R., Fakotakis, N. (eds.) SPECOM 2015. LNCS (LNAI), vol. 9319, pp. 25–33. Springer, Cham (2015). https://doi.org/10.1007/978-3-319-23132-7_3
Kingsbury, B.: Lattice-based optimization of sequence classification criteria for neural-network acoustic modeling. In: IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2009, pp. 3761–3764 (2009)
Ko, T., Peddinti, V., Povey, D., Khudanpur, S.: Audio augmentation for speech recognition. In: Proceedings of INTERSPEECH (2015)
Levin, K., et al.: Automated closed captioning for Russian live broadcasting. In: Proceedings of INTERSPEECH, pp. 1438–1442 (2014)
Liptchinsky, V., Synnaeve, G., Collobert, R.: Letter-based speech recognition with Gated ConvNets (2017). arxiv:1712.09444
Miao, Y., Gowayyed, M., Metze, F.: EESEN: end-to-end speech recognition using deep RNN models and WFST-based decoding. In: Proceedings of ASRU, pp. 167–174 (2015)
Povey, D., et al.: The Kaldi speech recognition toolkit. In: Proceedings of ASRU (2011)
Rumelhart, D.E., Hinton, G.E., Williams, R.J.: Neurocomputing: foundations of research. In: Learning Representations by Back-Propagating Errors, pp. 696–699. MIT Press, Cambridge (1988)
Zhou, Y., Xiong, C., Socher, R.: Improved regularization techniques for end-to-end speech recognition (2017). arxiv:1712.07108
Acknowledgements
This work was financially supported by the Ministry of Education and Science of the Russian Federation, Contract 14.575.21.0132 (IDRFMEFI57517X0132).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2018 Springer Nature Switzerland AG
About this paper
Cite this paper
Bataev, V., Korenevsky, M., Medennikov, I., Zatvornitskiy, A. (2018). Exploring End-to-End Techniques for Low-Resource Speech Recognition. In: Karpov, A., Jokisch, O., Potapova, R. (eds) Speech and Computer. SPECOM 2018. Lecture Notes in Computer Science(), vol 11096. Springer, Cham. https://doi.org/10.1007/978-3-319-99579-3_4
Download citation
DOI: https://doi.org/10.1007/978-3-319-99579-3_4
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-99578-6
Online ISBN: 978-3-319-99579-3
eBook Packages: Computer ScienceComputer Science (R0)