DSpace Repository

Selective EM Training of Acoustic Models based on Sufficient Statistics of Single Utterances

Show simple item record

dc.contributor.author Cincarek. Tobias en
dc.contributor.author Toda, Tomoki en
dc.contributor.author Saruwatari, Hiroshi en
dc.contributor.author Shikano, Kiyohiro en
dc.date.accessioned 2012-08-22T07:58:39Z en
dc.date.available 2012-08-22T07:58:39Z en
dc.date.issued 2005-11 en
dc.identifier.isbn 078039478X en
dc.identifier.uri http://hdl.handle.net/10061/7916 en
dc.description ASRU2005: IEEE Automatic Speech Recognition and Understanding Workshop, November 27, 2005, San Juan, Puerto Rico, US. en
dc.description.abstract In this paper, a new algorithm for selective training of acoustic models is proposed. The algorithm is formulated for an HMM-based model with Gaussian mixture densities, but works in principle for any statistical model, which has sufficient statistics. Since there are too many possibilities for selecting a data subset from a larger database, a heuristic has to be employed. The algorithm is based on deleting single utterances from a data pool temporarily or alternating between successive deletion or addition of utterances. The optimization criterion is the likelihood of the new model parameters given some development data, which can be calculated in a short amount of time based on sufficient statistics. The method is applied to automatically obtain task-dependent acoustic models for infant and elderly speech by selecting utterances from a data pool which are acoustically close to the development data. The proposed method is computationally practical and also addresses the issue of reducing the high costs evolving from the development of applications which make use of speech recognition technology en
dc.language.iso en en
dc.publisher IEEE en
dc.rights Copyright 2005 IEEE en
dc.title Selective EM Training of Acoustic Models based on Sufficient Statistics of Single Utterances en
dc.type.nii Conference Paper en
dc.textversion Publisher en
dc.identifier.spage 168 en
dc.identifier.epage 173 en
dc.relation.doi 10.1109/ASRU.2005.1566486 en
dc.identifier.NAIST-ID 73292716 en


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace


Advanced Search

Browse

My Account