MMSE-Based Missing-Feature Reconstruction With Temporal Modeling for Robust Speech Recognition

@article{Gonzlez2013MMSEBasedMR,
  title={MMSE-Based Missing-Feature Reconstruction With Temporal Modeling for Robust Speech Recognition},
  author={Jos{\'e} A. Gonz{\'a}lez and Antonio M. Peinado and Ning Ma and {\'A}ngel M. G{\'o}mez and Jon Barker},
  journal={IEEE Transactions on Audio, Speech, and Language Processing},
  year={2013},
  volume={21},
  pages={624-635}
}
This paper addresses the problem of feature compensation in the log-spectral domain by using the missing-data (MD) approach to noise robust speech recognition, that is, the log-spectral features can be either almost unaffected by noise or completely masked by it. First, a general MD framework based on minimum mean square error (MMSE) estimation is introduced which exploits the correlation across frequency bands to reconstruct the missing features. This framework allows the derivation of… CONTINUE READING