Web-derived Pronunciation Data

The following files contain the raw pronunciation data (prior to any correction or site-specific normalization) for English words extracted from non EU sites. The data is presented in a TAB-separated format, where the first column is the pronunciation, the second column is the orthography, and the rest of the columns contain information like the source webpage, extraction score, words in context, etc.

The non-EU IPA and Ad-hoc pronunciations contain two files named 'large' and 'small'. Note that 'small' is not a subset of 'large', but is extracted from a different set of websites [and is smaller in size :-)].

This data was extracted by Google from its web and news repositories for the 2008 Summer Research Workshop and is freely available to others under a Creative Commons Attribution 3.0 United States License.

	Ad-hoc data from non-EU sites	.tar.gz (195 MB)
	Ad-hoc data from news sites	.tar.gz (1.5 MB)
	IPA data from non-EU sites	.tar.gz (131 MB)

Citation

Arnab Ghoshal, Martin Jansche, Sanjeev Khudanpur, Michael Riley and Morgan Ulinski, "Web-derived pronunciations", in Proc. IEEE ICASSP, 2009. BIBTEX

Contact

For technical questions regarding the data, you can contact the team members listed below:

	Arnab Ghoshal	ag at jhu dot edu	Johns Hopkins University
	Martin Jansche	jansche at acm dot org	Google Inc
	Sanjeev Khudanpur	khudanpur at jhu dot edu	Johns Hopkins University
	Michael Riley	riley at google dot com	Google Inc
	Morgan Ulinski	meu3 at cornell dot edu	Cornell University

The page is maintained by Arnab Ghoshal.