Subspace-based speech enhancement by updating noise characteristics in the presence of speech

We present in this paper a signal subspace-based approach for enhancing a noisy signal. In our previous works we have developed an algorithm based on principal component analysis (PCA) in which the optimal subspace selection is provided by a variance of the reconstruction error (VRE) criterion. In this work we will improve our previous technique by applying an updating noise variance algorithm. The performance evaluations show that our method provides a higher noise reduction and a lower signal distortion than our previous one.