Simultaneous blind signal separation and denoising

This paper deals with the problem of blind separation of noisy signals. It proposes the application of the discrete wavelet transform as a preprocessing step of denoising prior to the blind separation algorithm. The separation of sinusoidal signals as well as speech signals in a noisy environment is studied with and without the use of the wavelet denoising step. The simulation results confirm the usefulness of this step. In all our simulation experiments, soft thresholding is used.