Speech Enhancement Using Sliding Window Empirical Mode Decomposition and Hurst-based Technique

The most challenging in speech enhancement technique is tracking non-stationary noises for long segments and low Signal-to-Noise Ratio (SNR). Different techniques have been proposed but, those were inaccurate highly noises. As a result, Empirical Mode Decomposition Hurst-based (EMDH) approach to enhance the signals corrupted by acoustic Hurst exponent statistics was adopted identifying selecting set of Intrinsic Functions (IMF) that are affected noise components. Moreover, signal reconstructed considering least IMF. Though it increases SNR, time resource consumption high. Also, requires significant improvement under nonstationary scenario. Hence, this article, EMDH enhanced using Sliding Window (SW) technique. In SWEMDH approach, computation EMD performed based on small sliding window along with axis. depends frequency band. possible discontinuities IMF between windows prevented total number modes sifting iterations should be priori. For each module, sifting iterations determined decomposition many standard algorithm calculating average steps module. Based complexity reduced significantly suitable quality decomposition. Finally, experimental results show considerable improvements environments.