The dimensionality and correlation between acoustic observation vectors and between the components within the vectors are investigated in terms of their impact on the performance of HMM (hidden Markov model)-based speech recognition. The dimensionality and correlation are manipulated with principal component analysis and linear discrimination analysis, on either a continuous density or a discrete density HMM system.
Keywords: HMM, Speech Recognition, PCA, LDA, VQ
Cite as: Wang, X., Bosch, L.F.M.t., Pols, L.C.W. (1993) Impact of dimensionality and correlation of observation vectors in HMM-based speech recognition. Proc. 3rd European Conference on Speech Communication and Technology (Eurospeech 1993), 1583-1586, doi: 10.21437/Eurospeech.1993-355
@inproceedings{wang93c_eurospeech, author={Xue Wang and Louis F. M. ten Bosch and Louis C. W. Pols}, title={{Impact of dimensionality and correlation of observation vectors in HMM-based speech recognition}}, year=1993, booktitle={Proc. 3rd European Conference on Speech Communication and Technology (Eurospeech 1993)}, pages={1583--1586}, doi={10.21437/Eurospeech.1993-355} }