Audio Declipping with (Weighted) Analysis Social Sparsity

TL;DR

This paper introduces an analysis (cosparse) approach to audio declipping, extending existing algorithms with weighted coefficients, and compares their performance, showing that weights can improve quality in some cases.

Contribution

It develops a novel analysis (cosparse) declipping algorithm with weighted coefficients and compares its performance to synthesis methods, highlighting efficiency and quality improvements.

Findings

Weights improve reconstruction in some cases

Analysis Empirical Wiener matches the quality of a more expensive method

Analysis variant with PEW slightly outperforms synthesis in auditory metrics

Abstract

We develop the analysis (cosparse) variant of the popular audio declipping algorithm of Siedenburg et al. (2014). Furthermore, we extend both the old and the new variants by the possibility of weighting the time-frequency coefficients. We examine the audio reconstruction performance of several combinations of weights and shrinkage operators. The weights are shown to improve the reconstruction quality in some cases; however, the best scores achieved by the non-weighted methods are not surpassed with the help of weights. Yet, the analysis Empirical Wiener (EW) shrinkage was able to reach the quality of a computationally more expensive competitor, the Persistent Empirical Wiener (PEW). Moreover, the proposed analysis variant incorporating PEW slightly outperforms the synthesis counterpart in terms of an auditorily motivated metric.