DSpace Repository

Spectral Subtraction in Noisy Environments Applied to Speaker Adaptation Based on HMM Sufficient Statistics

Show simple item record

dc.contributor.author Shingo Yamade ja
dc.contributor.author Kanako Matsunami ja
dc.contributor.author Akira Baba ja
dc.contributor.author Akinobu Lee ja
dc.contributor.author Hiroshi Saruwatari ja
dc.contributor.author Kiyohiro Shikano ja
dc.date.accessioned 2012-08-22T07:58:52Z
dc.date.available 2012-08-22T07:58:52Z
dc.date.issued 2002-09 ja
dc.identifier.uri http://hdl.handle.net/10061/8106
dc.description ICSLP2002: the 7th International Conference on Spoken Language Processing , September 16-20, 2002, Denver, Colorado, USA. ja
dc.description.abstract Noise and speaker adaptation techniques are essential to realize robust speech recognition in real noisy environments . In this paper, we applied spectral subtraction to an unsupervised speaker adaptation algorithm in noisy environments. The adaptation algorithm consists of the following five steps. (1) Spectral subtraction is carried out for noise added database. (2) Noise matched acoustic models are trained by using noise added speech database. (3) HMM sufficient statistics for each speaker are calculated from noise added speech database, and stored. (4) According to one arbitrary utterance, speakers close to a test speaker are selected by using speaker GMMs. (5) Speaker adapted acoustic models are constructed from HMM sufficient statistics of the selected speakers. We evaluated our unsupervised speaker adaptation algorithm in noisy environments in the 20k dictation task. The recognition experiments show that our speaker adapted acoustic model can achieve 82% word accuracy in 20dB SNR, which is about 6% higher than that of the noise matched models trained by Forward-Backward algorithm. ja
dc.language.iso en ja
dc.rights Copyright 2002 ISCA ja
dc.title Spectral Subtraction in Noisy Environments Applied to Speaker Adaptation Based on HMM Sufficient Statistics ja
dc.type.nii Conference Paper ja
dc.textversion Publisher ja
dc.identifier.spage 1045 ja
dc.identifier.epage 1048 ja


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace


Advanced Search

Browse

My Account