Абстрактный

Estimation of Clean Spectrogram Noisy Value Functions Based on Metropolis Iterative Algorithm.

Mahdi Jalali

The paper consisted of two parts. First, we estimated the clean speech signals from the estimated clean spectrograms with several values of K for one word. We then looked at the spectrograms of the estimated clean speech signals. Ideally, these two spectrograms (the estimated clean speech spectrogram and the spectrogram of the estimated clean speech) should be the same. We found that the spectrogram of the estimated clean speech signal with K=20 iterations looked closest to the estimated clean spectrogram. Next, we chose a column for which the estimated clean spectrogram and the spectrogram of the estimated clean speech signal visually differed