Speech Enhancement based on Wiener Filter and Compressive Sensing

In the last few decades, many advanced technologies have been proposed, in which communications played a great role as well as telecommunications applications. The noise elimination in various environments became the most concerned as it greatly hindered the speech communication applications. The improvement of noisy speech interms of quality and intelligibility are taken into account without introducingany additional noise. Many speech enhancement algorithms have been proposed. Wiener filter is one of the classical algorithm that improve the noisy speech by reducing its noise components through selectively chosen Wiener gain. In this paper, compressive sensing method by randomize measurement matrix is combined with the Wiener filter to reduce the noisy speech signal to produce high signal to noise ratio. The PESQ is used to measure the quality of the proposed algorithm design. Experimental results showthe effectiveness of our proposed algorithm to enhance noisy signals corrupted by various noises compared to other traditional algorithms, in which high PESQ scores were achieved across various noises and different SNRs.