Single channel adaptive noise cancelation for enhancing noisy speech

The paper presents a new single channel adaptive noise canceling technique for removing the harmful effects of additive noise on the speech signal. The conventional method makes a reference signal for adaptive filtering using the delay information, i.e., pitch period, estimated on a frame basis from the input speech. The proposed method, however, gets a properly delayed input speech on a sample basis as a reference signal using the recursion formula of the autocorrelation function or average magnitude difference function. By doing that, the problems of the conventional method such as discontinuity between frames, unsuitable processing of the unvoiced sound and the transition region from voiced to unvoiced sound or vice versa can be solved. Experimental results with normalized least mean square and fast least squares adaptive algorithms demonstrate that the proposed method improves the signal-to-noise ratios as well as the perceived speech quality better than the conventional method