Speech enhancement based on a hybrid a priori signal-to-noise ratio (SNR) estimator and a self-adaptive Lagrange multiplier

Speech enhancement techniques, using spectral subtraction, have the drawback of generating an annoying residual noise with musical character. An accurate estimate of the a priori SNR is critical for eliminating musical noise. In this paper, for an accurate estimate of the a priori SNR we have proposed a hybrid a priori SNR estimator and a self-adaptive Lagrange multiplier with Wiener denoising technique. Objective evaluations showed that the proposed method performed better than the Decision-Directed (DD) approach.