Audio Declipping by Cosparse Hard Thresholding

Recovery of clipped audio signals is a very challenging inverse problem. Recently, it has been successfully addressed by several methods based on the sparse synthesis data model. In this work we propose an algorithm for enhancement of clipped audio signals that exploits the sparse analysis (cosparse) data model. Experiments on real audio data indicate that the algorithm has better signal restoration performance than state-of-the-art sparse synthesis declipping methods.