Abstract
Norwegian is a challenging language for automatic speech recognition research because it has two written standards (Bokmål and Nynorsk) and a large number of distinct dialects, from which none has status of an official spoken norm. A traditional lexicon-based approach to ASR leads to a huge lexicon (because of the two standards and also due to compound words) with many spelling and pronunciation variants, and consequently to a large (and sparse) language model (LM). We have built a system with 601k-word lexicon and an acoustic model (AM) based on several types of neural networks and compare its performance with a lexicon-free end-to-end system developed in the ESPnet framework. For evaluation we use a publically available dataset of Norwegian parliament speeches that offers 100 h for training and 12 h for testing. In spite of this rather limited training resource, the lexicon-free approach yields significantly better results (13.0% word-error rate) compared to the best system with the lexicon, LM and neural network AM (that achieved 22.5% WER).
Keywords
- automatic speech recognition
- Norwegian
- Bokmål
- Nynorsk
- Deep neural network
- end-to-end speech recognition
This is a preview of subscription content, access via your institution.
Buying options
References
Frostad, B.H., Schall, V., Holte, S.M.: Towards ASR that supports linguistic diversity in Norway. In: Proceedings of LT4All, Paris, pp. 328–331 (2019)
Kvale, K.: Norwegian numerals: a challenge to automatic speech recognition. In: Proceedings of ICSLP, Philadelphia, pp. 2028–2031 (1996)
Johnsen, M. H., Amble, T., Harborg, E.: A norwegian spoken dialogue system for bus travel information. In: SLT in Telecommunications, pp. 125–131 (2003)
Markhus, V., Gajic, B., Svarverud, J., Solbraa, L. E., Johnsen, M. H.: Annotation and automatic recognition of spontaneously dictated medical records for Norwegian. In: Proceedings of NORSIG, pp. 316–319 (2004)
Adde, L.: A discriminative approach to pronunciation variation modeling in speech recognition. PhD Thesis, NTNU Trondheim (2013)
Kristensen, T. S., Nilssen, M. S.: Grapheme to phoneme conversion of norwegian using hidden markov models. In: 2020 15th IEEE Conference on Industrial Electronics and Applications (ICIEA), pp. 1121–1126 IEEE (2020)
Solberg, P. E., Ortiz, P.: The norwegian parliamentary speech corpus. arXiv preprint arXiv:2201.10881 (2022)
Nouza, J., Safarik, R., Cerva, P.: ASR for south slavic languages developed in almost automated way. In: Proceedings of Interspeech, pp. 3868–3872 (2016)
Safarik, R., Nouza, J. Unified approach to development of ASR systems for East Slavic languages. In: Proceedings of International Conference on Statistical Language and Speech Processing, pp. 193–203 (2017)
Nouza, J., Safarik, R.: Parliament archives used for automatic training of multi-lingual automatic speech recognition systems. In: Ekštein, K., Matoušek, V. (eds.) TSD 2017. LNCS (LNAI), vol. 10415, pp. 174–182. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-64206-2_20
Graves, A., Fernández, S., Gomez, F.: Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks. In: Proceedings of ICML, pp. 369–376 (2006)
Zhang, S., Jiang, H., Xiong, S., Wei, S., Dai, L.: Compact feedforward sequential memory networks for large vocabulary continuous speech recognition, In: Proceedings of Interspeech, San Francisco, pp. 3389–3393 (2016)
Graves, A., Jaitly, N.: Towards end-to-end speech recognition with recurrent neural networks. In: Proceedings of ICML, pp. 1764–1772 (2014)
Chorowski, J. K., Bahdanau, D., Serdyuk, D., Cho, K., Bengio, Y.: Attention-based models for speech recognition. In: Advances in Neural Information Processing Systems, vol. 28 (2015)
Watanabe, S., Hori, T., Karita, S., Hayashi, T., Nishitoba, J., Unno, N., et al.: ESPnet: end-to-end speech processing toolkit, In: Proceedings Interspeech, pp. 2207–2211 (2018)
Watanabe, S., Hori, T., Kim, S., Hershey, J.R., Hayashi, T.: Hybrid CTC/attention architecture for end-to-end speech recognition. IEEE J. Sel. Top. Sign. Process. 11(8), 1240–1253 (2017)
Sennrich, R., Haddow, B., Birch, A.: Neural machine translation of rare words with subword units. In: Proceedings of ACL, pp. 1715–1725 (2016)
Hannun, A. Sequence modeling with CTC, distill (2017). https://distill.pub/2017/ctc
Amodei, D., et al.: Deep speech 2: end-to-end speech recognition in English and Mandarin. In: Proceedings of PMLR, pp. 173–182 (2016)
Chen, G., et al.: Gigaspeech: an evolving, multi-domain ASR corpus with 10,000 hours of transcribed audio. In: Proceedings of Interspeech, pp. 3670–3674 (2021)
Acknowledgements
This work was supported by the Technology Agency of the Czech Republic (project No. TO01000027).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 Springer Nature Switzerland AG
About this paper
Cite this paper
Nouza, J., Červa, P., Žd’ánský, J. (2022). Lexicon-based vs. Lexicon-free ASR for Norwegian Parliament Speech Transcription. In: Sojka, P., Horák, A., Kopeček, I., Pala, K. (eds) Text, Speech, and Dialogue. TSD 2022. Lecture Notes in Computer Science(), vol 13502. Springer, Cham. https://doi.org/10.1007/978-3-031-16270-1_33
Download citation
DOI: https://doi.org/10.1007/978-3-031-16270-1_33
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-16269-5
Online ISBN: 978-3-031-16270-1
eBook Packages: Computer ScienceComputer Science (R0)