ISCA Archive Interspeech 2008
ISCA Archive Interspeech 2008

Improving speech systems built from very little data

John Kominek, Sameer Badaskar, Tanja Schultz, Alan W. Black

This paper studies two ways for helping non-specialist users develop speech systems from limited data for new languages. Focused web re-crawling finds additional examples of text matching the domain as specified by the user. This improves the language model and cuts word error rate nearly in half. Iterative voice building with interleaved lexicon construction uses the voice from a previous iteration to help construct an improved voice. 4.5 hours of the user's time reduces transcription error rate from 32% to 4%.

doi: 10.21437/Interspeech.2008-177

Cite as: Kominek, J., Badaskar, S., Schultz, T., Black, A.W. (2008) Improving speech systems built from very little data. Proc. Interspeech 2008, 1833-1836, doi: 10.21437/Interspeech.2008-177

  author={John Kominek and Sameer Badaskar and Tanja Schultz and Alan W. Black},
  title={{Improving speech systems built from very little data}},
  booktitle={Proc. Interspeech 2008},