Robustness of Linear Discriminant Analysis in Automatic Speech Recognitio

  title={Robustness of Linear Discriminant Analysis in Automatic Speech Recognitio},
  author={Marcel Katz and Hans-G{\"u}nter Meier and Hans J. G. A. Dolfing and Dietrich Klakow},
This paper focuseson the problemof a robust estimation of different transformationmatricesbasedon thewell knownlinear discriminantanalysis(LDA) as it is usedin automaticspeech recognition systems.We investigatethe effectof classdistributionswith artificial featuresandcompare theresultingFishercriterion. Thispapershowsthat it is notveryhelpfulto useonly theFishercriterion for anassessment of classseparability. Furthermore weaddressthe problemof dealingwith too manyadditionaldimensionsin… CONTINUE READING

From This Paper

Figures, tables, and topics from this paper.


Publications referenced by this paper.
Showing 1-2 of 2 references

Investigationof SiliconAuditoryModelsandGeneralization of Linear Discriminant Analysis for Improved Speec h Recognition, PhD Thesis,JohnHopkins

N. Kumar
View 3 Excerpts
Highly Influenced

LinearDiscriminantAnalysis for Improved Large Vacabulary ContinuousSpeechRecognition

R. Haeb-Umbach, andH. Ney
Proceedingsof the IEEE Int. Conf. on Acoustics, Speec h, andSignalProcessing, • 1992
View 2 Excerpts

Similar Papers

Loading similar papers…