This article proposes an efficient median filter based algorithm to remove transient noise in a speech signal. The\r\nproposed algorithm adopts a modified long-term predictor (LTP) as the pre-processor of the noise reduction\r\nprocess to reduce speech distortion caused by the nonlinear nature of the median filter. This article shows that the\r\nLTP analysis does not modify to the characteristic of transient noise during the speech modeling process.\r\nOppositely, if a short-term linear prediction (STP) filter is employed as a pre-processor, the enhanced output\r\nincludes residual noise because the STP analysis and synthesis process keeps and restores transient noise\r\ncomponents. To minimize residual noise and speech distortion after the transient noise reduction, a modified LTP\r\nmethod is proposed which estimates the characteristic of speech more accurately. By ignoring transient noise\r\npresence regions in the pitch lag detection step, the modified LTP successfully avoids being affected by transient\r\nnoise. A backward pitch prediction algorithm is also adopted to reduce speech distortion in the onset regions.\r\nExperimental results verify that the proposed system efficiently eliminates transient noise while preserving desired\r\nspeech signal.
Loading....