In this study, we propose an Adaptive Sub-Band Spectral Subtraction (ASBSS) method which can vary noise subtraction rate according to SNR in frequency bands at each frame. In the conventional Spectral Subtraction(SS), speech spectral is estimated by adjusting noise subtraction rate according to SNR. In general, SNR is defined and computed as the average over all the input speech signal. However...

The problem of spectral subtraction, to estimate the parameters of a single source in colored noise, is used to show the relationships between the likelihood formulation and spectral density estimation. Reported previously as a filter bank processing for spectral estimation, it is shown that the normalized Capon estimate is the natural tool for source location in I-d and 2-d scenarios when the ...

A speech enhancement system which utilizes an optimal (in the minimum mean square error sense) short-time spectral amplitude estimetor is described. The derivation of the optimal estimator is based on modeling speech es a quasi-periodic signal, and on applying spectral decomposition. The optimal spectral amplitude estimator and a recently developed vector spectral subtraction amplitude estimato...

In this paper we investigate the modulation domain as an alternative to the acoustic domain for speech enhancement. More specifically, we wish to determine how competitive the modulation domain is for spectral subtraction as compared to the acoustic domain. For this purpose, we extend the traditional analysis-modification-synthesis framework to include modulation domain processing. We then comp...

In a reverberant enclosure, acoustic speech signals are degraded by reflections from walls, ceilings, and objects. Restoring speech quality and intelligibility from reverberated speech has received increasing interest over the past few years. Although multiple channel dereverberation methods provide some improvements in speech quality/intelligibility, single-channel dereverberation remains an o...

In this paper, an enhancement method is proposed for a speech signal that is corrupted by background noise. The proposed method is based on the spectral subtraction technique. The conventional spectral subtraction method improves the quality of the signal by reducing noise but it introduces an annoying musical noise. To eliminate the musical noise, spectral subtraction method in modulation doma...

In this paper, we propose a method for enhancing speech corrupted by broadband noise. The method is based on the spectral subtraction technique. Besides reducing the noise conventional spectral subtraction introduces an annoying residual musical noise. To eliminate the musical noise we propose to introduce reduced varying scaling factor of spectral subtraction, with a following application of w...

In this paper we investigate the modulation domain as an alternative to the acoustic domain for speech enhancement. More specifically, we wish to determine how competitive the modulation domain is for spectral subtraction as compared to the acoustic domain. For this purpose, we extend the traditional analysis-modification-synthesis framework to include modulation domain processing. We then comp...

An auditory-motivated noise reduction methodology is proposed based on the inhibitory property of the human auditory system. The proposed psychoacoustic spectral subtraction method utilizes non-linearly and spatially distributed inhibitory weights which are determined based on the critical bands of the auditory system. Moreover, the weights are adjusted based on the segmental a posteriori SNR, ...