Minimum Bayes Risk decoding and system combination based on a recursion for edit distance

Authors:
Haihua Xu;Daniel Povey;Lidia Mangu;Jie Zhu
Affiliations:
Department of Electronic Engineering, Shanghai Jiao Tong University, Shanghai 200240, China;Microsoft Research, Redmond, WA, USA;IBM T.J. Watson Research Center, Yorktown Heights, NY, USA;Department of Electronic Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
Venue:
Computer Speech and Language
Year:
2011

Citing 4
Cited 0

BLEU: a method for automatic evaluation of machine translation

ACL '02 Proceedings of the 40th Annual Meeting on Association for Computational Linguistics
OpenFst: a general and efficient weighted finite-state transducer library

CIAA'07 Proceedings of the 12th international conference on Implementation and application of automata
The subspace Gaussian mixture model-A structured model for speech recognition

Computer Speech and Language
Tree-Based Covariance Modeling of Hidden Markov Models

IEEE Transactions on Audio, Speech, and Language Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Abstract: In this paper we describe a method that can be used for Minimum Bayes Risk (MBR) decoding for speech recognition. Our algorithm can take as input either a single lattice, or multiple lattices for system combination. It has similar functionality to the widely used Consensus method, but has a clearer theoretical basis and appears to give better results both for MBR decoding and system combination. Many different approximations have been described to solve the MBR decoding problem, which is very difficult from an optimization point of view. Our proposed method solves the problem through a novel forward-backward recursion on the lattice, not requiring time markings. We prove that our algorithm iteratively improves a bound on the Bayes risk.