Modified Wiener filtering speech enhancement algorithm with phase spectrum compensation
Zhang Wenlu, Hua Feng Peng · 2017
In this paper, a modified Wiener filtering speech enhancement algorithm with phase spectrum compensation is proposed, which aims at improving performance of typical Wiener filtering speech enhancement algorithm in low signal-noise ratio. Since typical speech enhancement algorithms always used the observed noisy speech phase spectrum unchanged directly as enhanced speech phase spectrum, and estimated noise power spectrum by the average noise power spectrum of the pauses in speech, so it is inaccurate for following signal processing. In order to solve these problems, the proposed method updates the Wiener filtering estimation of noise power spectrum according to whether the current frame to determine is speech signal or non-speech signal, and combines the amplitude estimation of clean speech signal with phase spectrum compensation, which retain more speech details information. Simulation results show the effectiveness of the proposed algorithm.