Perceptual Compressed Sensing and perceptual Sparse Fast Fourier Transform for audio signal compression

Abstract

Sparse signal processing has become central in many communications and multimedia processing applications. Techniques such as Compressed Sensing(CS) and Sparse Fast Fourier Transform (SFFT) provide good quality of the restored signal even when the signal is not completely sparse even also at high compression ratio. In this work, we examine the effect of taking the perceptual properties of the audio signal into account while performing both CS and SFFT. The improved perceptual signal quality by Mean Opinion Score (MOS) is verified.