ISCA Archive Interspeech 2008
ISCA Archive Interspeech 2008

HMM-based estimation of unreliable spectral components for noise robust speech recognition

Bengt J. Borgström, Abeer Alwan

This paper presents a novel approach for reconstructing unreliable spectral components, which utilizes HMM-based missing feature algorithms, and applies them to noise robust speech recognition. The proposed technique uses the forward-backward algorithm to estimate corrupt spectrographic data based on nearby reliable features, noisy observations, and on an underlying statistical model. The estimation process can be applied based on intra-channel information, intra-feature information, or a combination of both. The overall system is shown to provide vast improvements for the Consonant Challenge Database [1], for both MFCCs and PLP features, when using an oracle mask. Moreover, through downsampling of statistical models [2], the required complexity of the system is greatly reduced with negligible effects on results.

s http://www.odettes.dds.nl/challengeIS08/index.html B. J. Borgström and A. Alwan, An Efficient Approximation of the Forward-Backward Algorithm to Deal With Packet Loss, With Applications to Remote Speech Recognition, Proc. of ICASSP 2008.


doi: 10.21437/Interspeech.2008-487

Cite as: Borgström, B.J., Alwan, A. (2008) HMM-based estimation of unreliable spectral components for noise robust speech recognition. Proc. Interspeech 2008, 1769-1772, doi: 10.21437/Interspeech.2008-487

@inproceedings{borgstrom08_interspeech,
  author={Bengt J. Borgström and Abeer Alwan},
  title={{HMM-based estimation of unreliable spectral components for noise robust speech recognition}},
  year=2008,
  booktitle={Proc. Interspeech 2008},
  pages={1769--1772},
  doi={10.21437/Interspeech.2008-487}
}