Speech enhancement using improved MAP estimation and Wiener filter

In single channel speech enhancement using Wiener filters, the accurate estimation of speech and noise periodograms forming the noisy speech periodogram is really important since it highly affects the performance. In some methods a Gaussian Mixture Model (GMM) is used to model the Probability Density Function (PDF) of speech and noise periodograms and then a Maximum A-Posteriori criterion is applied on the PDFs to estimate these periodograms. These MAP methods exhibited very good enhancement results. A mathematical improvement on the extraction of the MAP estimation of the speech periodogram is introduced here. This improvement is based on the estimation of spectrum amplitude and then using it for periodogram estimation. Our method exhibits good performance results in comparison to other successful methods.