Speech Enhancement Based on a Combination of Spectral Subtraction and a Minimum Mean-Square Error Short-Time Log-Spectral Amplitude Estimator in Wavelet Domain

摘要

The corruption of speech due to presence of additive background noise causes severe difficulties in various communication environments. This paper presents a novel noise reduction technique based upon a combination of cascaded spectral subtraction and minimum mean-square error short-time log-spectral amplitude estimator methods in the wavelet domain. The scheme's performance is illustrated by experiments in a noisy car environment, in comparison with spectral subtraction and a minimum mean-square error short-time log-spectral amplitude estimator.