Post-filter for multichannel malay speech enhancement in various indoor noisy environments

Noise suppression is vital in audio processing for improving speech clarity and quality, particularly in indoor environments where reverberations and various noise sources make it difficult to maintain signal clarity. This research examines the performance of the Relative Transfer Matrix (ReTM) using different post-filters in various distances of noise speech sources to tackle the challenges posed by indoor acoustics. ReTM is applied to handle situations with multiple sound sources, capturing complex spatial relationships and improving the noise reduction performance. The experiment used the Malaya-Speech datasets with noise samples from the DEMAND database. The findings indicate notable enhancements in speech quality and intelligibility across various noisy environments and speaker locations, underlining the promise of the post-filters for practical application.