ISCA Archive Interspeech 2005
ISCA Archive Interspeech 2005

Confidence scoring and rejection using multi-pass speech recognition

Vincent Vanhoucke

This paper presents a computationally efficient method for using multiple speech recognizers in a multi-pass framework to improve the rejection performance of an automatic speech recognition system. A set of criteria is proposed, which determine at run time when rescoring using a second pass is expected to improve the rejection performance. The second pass result is used along with a set of features derived from the first pass to compute a combined confidence score. The feature combination is optimized globally based on training data. The combined system significantly outperforms a simple two-pass system at little more computational cost than comparable one-pass and two-pass systems.

doi: 10.21437/Interspeech.2005-266

Cite as: Vanhoucke, V. (2005) Confidence scoring and rejection using multi-pass speech recognition. Proc. Interspeech 2005, 3133-3136, doi: 10.21437/Interspeech.2005-266

  author={Vincent Vanhoucke},
  title={{Confidence scoring and rejection using multi-pass speech recognition}},
  booktitle={Proc. Interspeech 2005},