Open AccessDifferences Amplitude Based Improvement of Minimum Variance Distortionless Response Filter
Author(s) -
Quan Trong The
Publication year - 2020
Publication title -
international journal of recent technology and engineering
Language(s) - English
Resource type - Journals
ISSN - 2277-3878
DOI - 10.35940/ijrte.e6151.018520
Subject(s) - filter (signal processing) , speech enhancement , noise (video) , computer science , amplitude , a priori and a posteriori , minimum variance unbiased estimator , interference (communication) , signal (programming language) , speech recognition , variance (accounting) , filter design , mathematics , algorithm , statistics , telecommunications , artificial intelligence , physics , mean squared error , computer vision , philosophy , channel (broadcasting) , accounting , epistemology , quantum mechanics , business , image (mathematics) , programming language
In this work, the author introduces a new technique for improving the performance of minimum variance distortionless response filter in condition of coherent noise. The proposal algorithm exploits a priori information of differences amplitude to balance power spectral densities of observed noisy signals. The output signal of MVDR filter is then processed by an additional post-filtering, which based speech presence probability to suppress more noise interference and increase quality speech. In experiments using two noisy signal recordings in anechoeic room, the modified MVDR-filter results provides that the suggested algorithm increases speech quality compared to the conventional MVDR filter.