Iterative Decoding: A Novel Re-scoring Framework for Confusion Networks

Anoop Deoras and Fred Jelinek

Abstract

Recently there has been a lot of interest in confusion network re-scoring using

sophisticated and complex knowledge sources. Traditionally, re-scoring has been carried

out by the N-best list method or by the lattices or confusion network dynamic programming

method. Although the dynamic programming method is optimal, it allows for the

incorporation of only Markov knowledge sources. N-best lists, on the other hand, can

incorporate sentence level knowledge sources, but with increasing N, the re-scoring

becomes computationally very intensive. In this paper, we present an elegant framework

for confusion network re-scoring called ’Iterative Decoding’. In it, integration of multiple

and complex knowledge sources is not only easier but it also allows for much faster re-

scoring as compared to the N-best list method. Experiments with Language Model re-scoring

show that for comparable performance (in terms of word error rate (WER)) of Iterative

Decoding and N-best list re-scoring, the search effort required by our method is 22 times

less than that of the N-best list method.

Details

Publication typeInproceedings
URLhttps://www.youtube.com/watch?v=iMYI41jfwA8
PublisherIEEE Workshop on Automatic Speech Recognition and Understanding
> Publications > Iterative Decoding: A Novel Re-scoring Framework for Confusion Networks