Mova0.5 pre-release⌕
☰
home / docs / papers / Semitic languages / Transfer Learning for Less-Resourced Semitic Languages Speech Recognition: the Case of Amharic

Transfer Learning for Less-Resourced Semitic Languages Speech Recognition: the Case of Amharic

What it is about

Amharic has little speech data, so the author reuses acoustic models trained on English (100 h and 460 h) and Mandarin (178 h). He keeps the hidden layers of each source network, replaces the output layer, and fine-tunes on 20 h of Amharic read speech with a smaller learning rate. Word error rate on Amharic falls from 38.72% (baseline) to 24.50% with the English-460 h source (14.22 points absolute), to 25.06% with English-100 h, and to 28.47% with Mandarin. The author judges the relatedness of source and target to matter more than the other factors he tests (source model quality and source data size). He names the Amharic lexicon (65k entries, against 130k for English), the letter-to-sound rules, and an epenthetic vowel that is spoken but not written as the main error sources.

Ideas

For Mova

I1 goes into tools/hbo (segmenter tables and a held-out OOV count). I2 is a rule table for Phonology of Biblical Hebrew. I3 is a test across the varieties in languages/arc/. Mova does no speech; only the text-side ideas apply.

Cite

The paper itself, not this card. BibTeX from the library's record of the work.

@inproceedings{woldemariam2020transfer,
  title = {{Transfer Learning for Less-Resourced Semitic Languages Speech Recognition: the Case of Amharic}},
  author = {Woldemariam, Yonas},
  booktitle = {Proceedings of the 1st Joint Workshop on Spoken Language Technologies for Under-resourced languages (SLTU) and Collaboration and Computing for Under-Resourced Languages (CCURL)},
  year = {2020},
  url = {https://aclanthology.org/2020.sltu-1.9/}
}

References

43 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Abate, S., Tachbelie, M., and Menze, W. (2009). Amharic speech recognition: Past present and future. In In: Proceedings of the 16th International Conference of Ethiopian Studies, pages 1391–1401. find
  2. Besacier, L., Barnard, E., Karpov, A., and Schultz, T. (2014). Automatic speech recognition for under-resourced languages. Speech Communication, 56:85– 100. find
  3. Bu, H., Du, J., Na, X., Wu, B., and Zheng, H. (2017). Aishell-1: An open-source mandarin speech corpus and a speech recognition baseline. 2017 20th Conference of the Oriental Chapter of the International Coordinating Committee on Speech Databases and Speech I/O Systems and Assessment (O-COCOSDA), pages 1–5. find
  4. Chang, S.-F., Manmatha, R., and Chua, T.-S. (2005). Combining text and audio-visible features in video indexing. In Acoustics, Speech, and Signal Processing, pages 1005–1008. find
  5. Das, A., Jyothi, P., and Hasegawa-Johnson, M. (2016). Automatic speech recognition using probabilistic transcriptions in swahili, amharic, and dinka. In INTER-SPEECH, pages 3524–3528. find
  6. Demeke, Y. and Hailemariam, S. (2012). Duration modeling of phonemes for amharic text to speech system. In Proceedings of the International Conference on Management of Emergent Digital EcoSystems (MEDES), pages 1–7. find
  7. Durrett, G., Pauls, A., and Klein, D. (2012). Syntactic transfer using a bilingual lexicon. In EMNLP-CoNLL. find
  8. Elmahdy, M., Hasegawa-Johnson, M., and Mustafawi, E. (2013). A transfer learning approach for under-resourced arabic dialects speech recognition. In Proceedings of the 6th Language and Technology Conference, page 290â293. find
  9. Farzad, E. and Eva, K. (1998). Speech technology in computer-aided language learning: Strengths and limitations of a new call paradigm. Language Learning & Technology, 2(1):45–60. find
  10. Feng, S. and Lee, T. (2018). Improving cross-lingual knowledge transferability using multilingual tdnn-blstm with language-dependent pre-final layer. In Interspeech, pages 2439–2443. find
  11. Gales, M. (1998). Maximum likelihood linear transformations for hmm-based speech recognition. Computer Science and Language, 12:75–98. find
  12. Gashaw, A. (2017). Rhythm in ethiopian english: Implications for the teaching of english prosody. International Journal of Education and Literacy Studies, 5(1):13–19. find
  13. George, S., Hagen, S., David, N., and Michael, P. (2013). Speaker adaptation of neural network acoustic models using i-vectors. In 2013 IEEE Workshop on Automatic Speech Recognition and Understanding, pages 55–59. find
  14. Ghahremani, P., Manohar, V., Hadian, H., Povey, D., and Khudanpur, S. (2017). Investigation of transfer learning for asr using lf-mmi trained neural networks. 2017 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), pages 279–286. find
  15. Ghoshal, A., Swietojanski, P., and Renals, S. (2013). Multilingual training of deep neural networks. 2013 IEEE International Conference on Acoustics Speech and Signal Processing, pages 7319–7323. find
  16. Gopinath, R. (1998). Maximum likelihood modeling with gaussian distributions for classification. In Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP’98, pages 661– 664. find
  17. Hailu, N. and Hailemariam, S. (2012). Modeling improved syllabification algorithm for amharic. In Proceedings of the International Conference on Management of Emergent Digital EcoSystems, pages 16–21. find
  18. Huang, J.-T., Li, J., Yu, D., Deng, L., and Gong, Y. (2013). Cross-language knowledge transfer using multilingual deep neural network with shared hidden layers. 2013 IEEE International Conference on Acoustics Speech and Signal Processing, pages 7304–7308. find
  19. Judith, R., Rotem, K., and MyiLibrary. (2008). Globally speaking : motives for adopting English vocabulary in other languages. Clevedon UK ; Buffalo [N.Y.] : Multilingual Matters, New York. find
  20. Karafiát, M., Baskar, M. K., Matejka, P., Veselý, K., Grézl, F., Burget, L., and Cernocký, J. (2017). 2016 but babel system: Multilingual blstm acoustic model with ivector based adaptation. In The Proceedings of INTER-SPEECH 2017, pages 719–723. find
  21. Karafiát, M., Baskar, M. K., Veselý, K., Grézl, F., Burget, L., and ernocký, J. (2018). Analysis of multilingual blstm acoustic model on low and high resource languages. 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 5789–5793. find
  22. Le, N., Bredin, H., Sargent, G., India, M., Lopez-Otero, P., Barras, C., Guinaudeau, C., Gravier, G., da Fonseca, G. B., Freire, I. L., do Patrocı́nio, Z. K. G., Guimarães, S. J. F., Martı́, G., Morros, J. R., Hernando, J., Fernández, L. D., Garcı́a-Mateo, C., Meignier, S., and Odobez, J.-M. (2017). Towards large scale multimedia indexing: A case study on person discovery in broadcast news. In Proceedings of International Workshop on Content-Based Multimedia Retrieval, pages 1–6. find
  23. Manohar, V., Povey, D., and Khudanpur, S. (2017). Jhu kaldi system for arabic mgb-3 asr challenge using diarization audio-transcript alignment and transfer learning. 2017 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), pages 346–352. find
  24. Melese, M., Besacier, L., and Meshesha, M. (2017). Amharic-english speech translation in tourism domain. In SCNLP@EMNLP 2017. find
  25. Mezaris, V., Gidaros, S., Papadopoulos, G. T., Kasper, W., Steffen, J., Ordelman, R., Huijbregts, M., de Jong, F., Kompatsiaris, Y., and Strintzis, M. G. (2010). A system for the semantic multimodal analysis of news audio-visual content. EURASIP Journal on Advances in Signal Processing, 2010:1–16. find
  26. Min, L. and Shrikanth, N. (2005). Toward detecting emotions in spoken dialogs. IEEE transactions on speech and audio processing, 13(2):293–303. find
  27. Peddinti, V., Povey, D., and Khudanpur, S. (2015). A time delay neural network architecture for efficient modeling of long temporal contexts. In The Proceedings of IN-TERSPEECH 2015, pages 3214–3218. find
  28. Povey, D., Ghoshal, A., Boulianne, G., Burget, L., Glembek, O., Goel, N. K., Hannemann, M., Motlı́cek, P., Qian, Y., Schwarz, P., Silovský, J., Stemmer, G., and Veselý, K. (2011). The Kaldi speech recognition toolkit. In IEEE 2011 Workshop on Automatic Speech Recognition and Understanding. find
  29. Rabiner, L. R. (1989). A tutorial on hidden Markov models and selected applications in speech recognition. Proceedings of the IEEE, 77(2):257–286. find
  30. Rosenberg, A., Audhkhasi, K., Sethy, A., Ramabhadran, B., and Picheny, M. (2017). End-to-end speech recognition and keyword search on low-resource languages. 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 5280–5284. find
  31. Sebsibe, H., Prahallad, K., Alan, B., Rohit, K., and Rajeev, S. (2004). Unit selection voice for amharic using festvox. In Fifth ISCA Workshop on Speech Synthesis, pages 103–107. find
  32. Tachbelie, M., Abate, S., and Besacier, L. (2014). Using different acoustic lexical and language modeling units for asr of an under-resourced language - amharic. Speech Communication, 56:181–194. find
  33. Wang, D. and Zheng, T. F. (2015). Transfer learning for speech and language processing. 2015 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA), pages 1225–1237. find
  34. Woldemariam, Y. (2018). Natural Language Processing in Cross-Media Analysis. Licentiate thesis, Faculty of Science and Technology, Umeå University, Jun. find
  35. Yan, J., Yu, H., and Li, G. (2018). Tibetan acoustic model research based on tdnn. 2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), pages 601–604. find
  36. Yifiru, M. (2003). Automatic amharic speech recognition system to command and control computers. Master’s thesis, School of Information Studies for Africa, Addis Ababa University. find
  37. Yimam, B. (2000). Amharic Grammar. Eleni Publishing Ltd, Addis Ababa. find
  38. Zhuang, X., Ghoshal, A., Rosti, A.-V., Paulik, M., and Liu, D. (2017). Improving dnn bluetooth narrowband acoustic models by cross-bandwidth and cross-lingual initialization. In INTERSPEECH, pages 2148–2152. 10. find
  39. Language Resource References Hui Bu and Jiayu Du and Xingyu Na and Bengu Wu and Hao Zheng. (2017). Aishell ASR corpus. provided by Beijing Shell Technology and distributed via OpenSLR, ISLRN SLR33. find
  40. ELRA-W0074. (2014). Amharic-English bilingual corpus, distributed via ELRA,1.0. distributed via ELRA,1.0, 1.0, ISLRN 590-255-335-719-0. find
  41. Elodie Gauthier and Laurent Besacier and Sylvie Voisin and Michael Melese and Uriel Pascal Elingui. (2016). Collecting Resources in Sub-Saharan African Languages for Automatic Speech Recognition: a Case Study of Wolof. European Language Resources Association (ELRA), Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC’16). find
  42. HaBiT. (2016). Harvesting big text data for under-resourced languages. distributed via Natural Language Processing Centre, Faculty of Informatics, Masaryk University. find
  43. Vassil Panayotov and Guoguo Chen and Daniel Povey and Sanjeev Khudanpur. (2015). LibriSpeech ASR corpus. distributed via OpenSLR, ISLRN SLR12. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2020-woldemariam-transfer-learning-less-resourced-semitic/