Speech Recognition with Segmental Conditional Random Fields: A Summary of the JHU CLSP 2010 Summer Workshop

  • Geoffrey Zweig ,
  • Patrick Nguyen ,
  • Dirk Van Compernolle ,
  • Kris Demuynck ,
  • Les Atlas ,
  • Pascal Clark ,
  • Greg Sell ,
  • Meihong Wang ,
  • Fei Sha ,
  • Hynek Hermansky ,
  • Damianos Karakos ,
  • Aren Jansen ,
  • Samuel Thomas ,
  • Samuel Bowman ,
  • Justine Kao ,
  • G.S.V.S. Sivaram

ICASSP 2011 |

Published by IEEE

Download BibTex

This paper summarizes the 2010 CLSP Summer Workshop on speech recognition at Johns Hopkins University. The key theme of the workshop was to improve on state-of-the-art speech recognition systems by using Segmental Conditional Random Fields (SCRFs) to integrate multiple types of information. This approach uses a state-of-the-art baseline as a springboard from which to add a suite of novel features including ones derived from acoustic templates, deep neural net phoneme detections, duration models, modulation features, and whole word point-process models. The SCRF framework is able to appropriately weight these different information sources to produce significant gains on both the Broadcast News and Wall Street Journal tasks.