Single‐Channel Noise Reduction

Abstract

This chapter is concerned with algorithms for reducing additive noise when only a single microphone signal is available. It provides an introduction to the basic principles and implementation aspects. The chapter introduces the Wiener filter and “spectral subtraction,” as well as noise power spectral density estimation techniques. It then focuses on non-linear optimal estimators. These estimators explicitly use the probability density function of short-time spectral coefficients and are thus better able to utilize the information available a priori. The chapter also presents maximum likelihood, maximum a posteriori, and minimum mean square estimators for the estimation of the complex discrete Fourier transform coefficients, as well as for functions of the spectral amplitude. It concludes with an overview of speech enhancement approaches using deep neural networks.

Authors 2

1. Peter Vary Aachen Institute of Communication Systems

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University, Institute of Communication Systems, Aachen, Germany

2. Rainer Martin

   Ruhr University Bochum

   Affiliation as printed

   Ruhr-Universität Bochum, Institute of Communication Acoustics, Bochum, Germany