Empirical Investigation of Compressed Sensing applicability to Lossy Audio Compression

Compressive sampling is a new framework that exploits sparsity of a signal in a transform domain to perform sampling below the Nyquist rate. In this paper we investigate the applicability of the Compressed Sensing Framework to audio compression by searching for a good sparsity basis and a reconstruction technique fit to audio applications. We also propose a new method for lossy audio compression of real, non-sparse audio signals, based on our investigations. The method uses the Modified Discrete Cosine Transform (MDCT) as a sparse basis and the l-1 norm optimization for signal reconstruction. We evaluate final audio quality with the Perceptual Evaluation of Audio Quality (PEAQ) algorithm. The method we propose has the properties of reverse-complexity, cryptography, error-resiliency and universality of encoder, altogether without any additional hardware.