Block-wise incremental adaptation algorithm for maximum kurtosis beamforming
Authors:
- 1 October 2011
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 19311168,p. 229-232
Abstract
In prior work, the current authors investigated beamforming algorithms that exploit the non-Gaussianity of human speech. The beamformers proposed in [1, 2, 3] are designed to maximize the kurtosis or negentropy of the subband output subject to the distortionless constraint for the direction of interest. Such techniques are able to suppress interference signals as well as reverberation effects without signal cancellation. They require, however, multiple passes of processing for each utterance in order to estimate the active weight vector. Hence, they are unsuitable for online implementation. In this work, we propose an online implementation of the maximum kurtosis beamformer. In a set of distant speech recognition experiments on far-field data, we demonstrate the effectiveness of the proposed technique. Compared to a single channel of the array, the proposed algorithm reduced word error rate from 15.4% to 6.5%.Keywords
This publication has 14 references indexed in Scilit:
- On the combination of voice prompt suppression with maximum kurtosis beamforming2011
- Adaptive Beamforming with a Maximum Negentropy Criterion2008
- Filter bank design based on minimization of individual aliasing terms for minimum mutual information subband adaptive beamforming2008
- A parametric approach to vocal tract length normalization2002
- Independent component analysis: algorithms and applicationsNeural Networks, 2000
- Semi-tied covariance matrices for hidden Markov modelsIEEE Transactions on Speech and Audio Processing, 1999
- Analysis of noise reduction and dereverberation techniques based on microphone arrays with postfilteringIEEE Transactions on Speech and Audio Processing, 1998
- Maximum likelihood linear transformations for HMM-based speech recognitionComputer Speech & Language, 1998
- Maximum likelihood linear regression for speaker adaptation of continuous density hidden Markov modelsComputer Speech & Language, 1995
- A Threshold Selection Method from Gray-Level HistogramsIEEE Transactions on Systems, Man, and Cybernetics, 1979