A wavelet transform-based blind adaptive filter of unknown noise from speech

This paper is concerned with the application of the wavelet transform (WT) in blind adaptive filtering (BAF) of speech signals under the single microphone constraint when neither speech nor noise are separately accessible. The paper describes the BAF system which performs the WT of a noisy speech signal and identifies and separates the noise from the speech. The nonstationarity property of human speech along with the estimated speech and noise energy over time and the estimated AR/ARMA parameters of WT coefficients of the input signal are utilized to construct cost functionals. The minimization of these cost functionals tunes sets of weights that multiply WT coefficients on different scales of the WT to produce a significantly improved signal to noise ratio and intelligibility of the filtered speech signal at the output of the BAF system.