An Improvement of Minimum Variance Distortionless Response Filter

In this paper, the author introduces an improvement of Minimum Variance Distortionless Response’s performance, which uses a priori information of speech presence probability to estimate the necessary matrix of noise. The proposal algorithm computes the smoothing parameter, that adapts with the presence or absence of speech components. The significant amount of noise reduction has provided the effectiveness and ability of increasing the signal-to-noise ratio of this algorithm’s speech enhancement. Post-filtering is an additional technique to enhance the quality of the output signal. The evaluation is presented in promising results of amplitude, spectrogram of original and processed signals.