Interspeech'2005 - Eurospeech
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.
Bibliographic reference. Vanhoucke, Vincent (2005): "Confidence scoring and rejection using multi-pass speech recognition", In INTERSPEECH-2005, 3133-3136.