Getting More Mileage from Web Text Sources for Conversational Speech Language Modeling using Class-dependent Mixtures

@inproceedings{citeulike:3198180, abstract = {Sources of training data suitable for language modeling of conversational speech are limited. In this paper, we show how training data can be supplemented with text from the web filtered to match the style and/or topic of the target recognition task, but also that it is possible to get bigger performance gains from the data by using class-dependent interpolation of N-grams. 1}, author = {Bulyko, Ivan and Ostendorf, Mari}, booktitle = {Proc. HLT-NAACL 2003}, citeulike-article-id = {3198180}, keywords = {language, model, text, web}, pages = {7--9}, posted-at = {2008-09-05 23:41:50}, priority = {0}, title = {Getting More Mileage from Web Text Sources for Conversational Speech Language Modeling using Class-dependent Mixtures}, url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.13.8175}, volume = {2}, year = {2003} }

TY - CONF ID - citeulike:3198180 L3 - citeulike-article-id:3198180 N2 - Sources of training data suitable for language modeling of conversational speech are limited. In this paper, we show how training data can be supplemented with text from the web filtered to match the style and/or topic of the target recognition task, but also that it is possible to get bigger performance gains from the data by using class-dependent interpolation of N-grams. 1 EP - 9 TI - Getting More Mileage from Web Text Sources for Conversational Speech Language Modeling using Class-dependent Mixtures VL - 2 T2 - Proc. HLT-NAACL 2003 SP - 7 KW - language KW - model KW - text KW - web AU - Bulyko, Ivan AU - Ostendorf, Mari PY - 2003/// UR - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.13.8175 ER -