Constrained Iterative Speech Enhance with Application to Automatic Speech Recognition
John H. L. Hansen, Mark A. Clements · 1988
set of iterative speech enhancement techniques employing specral constraints is extended and evaluated in this paper. The orignal unconstrained technique attempts to solve for the maximum ikelihood estimate of a speech waveform in additive noise. The new approaches (presented in ICASSP-87 [3]), apply inter- and intraframe spectral constraints to ensure optimum speech quality across all classes of speech. Constraints are applied based on the presence of perceptually important speech characteristics found during the echnique is presented the techniques have colored noise. And d to determine their extremely noisy en