Speech Enhancement Based on Nonlinear Spectral Subtraction.

Speech Enhancement Based on Nonlinear Spectral Subtraction.

Paper in proceedings (conference paper)

The spectral subtraction method is the most suitable technique for the elimination of stationary noise from degraded speech signal. In the following paragraphs the basic principles of nonlinear spectral subtraction are described, which optimize the subtractive factor in dependence on input-degraded speech signal. In this way the undesirable musical noise remainder can be suppressed, which arises after spectral subtraction. The results obtained are computed by the standard spectral subtraction method.

The spectral subtraction method is the most suitable technique for the elimination of stationary noise from degraded speech signal. In the following paragraphs the basic principles of nonlinear spectral subtraction are described, which optimize the subtractive factor in dependence on input-degraded speech signal. In this way the undesirable musical noise remainder can be suppressed, which arises after spectral subtraction. The results obtained are computed by the standard spectral subtraction method.

single channel, subtractive-type algorithms, speech detection, noise reduction

single channel, subtractive-type algorithms, speech detection, noise reduction

PORUBA, J.

2011

01.01.2002

Aruba

0-7803-7381-2

Proceedings of 2002 Fourth IEEE International Caracas Conference on Devices, Circuits and Systems

T031-1

4