Improved subword modeling for WFST-based speech recognition

Peter Smit, Sami Virpioja, Mikko Kurimo

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

26 Citations (Scopus)
845 Downloads (Pure)


Because in agglutinative languages the number of observed word forms is very high, subword units are often utilized in speech recognition. However, the proper use of subword units requires careful consideration of details such as silence modeling, position-dependent phones, and combination of the units. In this paper, we implement subword modeling in the Kaldi toolkit by creating modified lexicon by finite-state transducers to represent the subword units correctly. We experiment with multiple types of word boundary markers and achieve the best results by adding a marker to the left or right side of a subword unit whenever it is not preceded or followed by a word boundary, respectively. We also compare three different toolkits that provide data-driven subword segmentations. In our experiments on a variety of Finnish and Estonian datasets, the best subword models do outperform word-based models and naive subword implementations.
The largest relative reduction in WER is a 23% over word-based models for a Finnish read speech dataset. The results are also better than any previously published ones for the same datasets, and the improvement on all datasets is more than 5%.
Original languageEnglish
Title of host publicationProceedings of Interspeech 2017
PublisherInternational Speech Communication Association
Number of pages5
ISBN (Print)978-1-5108-4876-4
Publication statusPublished - Aug 2017
MoE publication typeA4 Article in a conference publication
Duration: 1 Jan 1900 → …

Publication series

NameInterspeech: Annual Conference of the International Speech Communication Association
ISSN (Electronic)1990-9772


Period01/01/1900 → …


  • speech recognition
  • Kaldi
  • subword modeling
  • Finnish
  • Estonian

Fingerprint Dive into the research topics of 'Improved subword modeling for WFST-based speech recognition'. Together they form a unique fingerprint.

Cite this