A maximum entropy language model integrating N-grams and topic dependencies for conversational speech recognition

Authors:
S. Khudanpur;Jun Wu
Affiliations:
Center for Language & Speech Process., Johns Hopkins Univ., Baltimore, MD, USA;-
Venue:
ICASSP '99 Proceedings of the Acoustics, Speech, and Signal Processing, 1999. on 1999 IEEE International Conference - Volume 01
Year:
1999

Citing 0
Cited 7

Topic-based mixture language modelling

Natural Language Engineering
Resource selection for domain-specific cross-lingual IR

Proceedings of the 27th annual international ACM SIGIR conference on Research and development in information retrieval
Latent semantic information in maximum entropy language models for conversational speech recognition

NAACL '03 Proceedings of the 2003 Conference of the North American Chapter of the Association for Computational Linguistics on Human Language Technology - Volume 1
Maximum Entropy Modeling: A Suitable Framework to Learn Context-Dependent Lexicon Models for Statistical Machine Translation

Machine Learning
Understanding without formality: augmenting speech recognition to understand informal verbal commands

Proceedings of the 43rd annual Southeast regional conference - Volume 1
Domain N-gram construction and its application to text editor

ACIIDS'11 Proceedings of the Third international conference on Intelligent information and database systems - Volume Part I
Are morphosyntactic taggers suitable to improve automatic transcription?

TSD'06 Proceedings of the 9th international conference on Text, Speech and Dialogue

Quantified Score

Hi-index	0.00

Visualization

Abstract

A compact language model which incorporates local dependencies in the form of N-grams and long distance dependencies through dynamic topic conditional constraints is presented. These constraints are integrated using the maximum entropy principle. Issues in assigning a topic to a test utterance are investigated. Recognition results on the Switchboard corpus are presented showing that with a very small increase in the number of model parameters, reduction in word error rate and language model perplexity are achieved over trigram models. Some analysis follows, demonstrating that the gains are even larger on content-bearing words. The results are compared with those obtained by interpolating topic-independent and topic-specific N-gram models. The framework presented here extends easily to incorporate other forms of statistical dependencies such as syntactic word-pair relationships or hierarchical topic constraints.