AN IMPROVED (AUTO-I, LSP-T) CONSTRAINED ITERATIVE SPEECH ENHANCEMENT FOR COLORED NOISE ENVIRONMENTS

In this correspondence we illustrate hom the (Auto:I, LSP:T) constrain ed iterative speech enhancement algorithm can be extended to provide i mproved performance in colored noise environments. The modified algori thm, referred to here as noise adaptive (Auto:I, LSP:T), operates on s ubbanded signal components in which the terminating iteration is adjus ted based on the a posteriori estimate of the signal-to-noise ratio (S NR) in each signal subband. The enhanced speech is formulated as a com bined estimate from individual signal subband estimators. The algorith m is shown to improve objective speech quality in additive noise envir onments over the traditional constrained iterative (Auto:I, LSP:T) enh ancement formulation.