We present a flexible dual channel noise reduction algorithm that combines the information of two microphone channels in the discrete Fourier transform domain. This algorithm can cope with substantially different signal-to-noise ratios at different time-frequency bins in both channels. The output is obtained by combining the outputs of two single channel filters and a dual channel filter weighted by the probability of speech presence. The algorithm has a low latency and does not require any additional information such as the microphone positions.
Cite as: Gerkmann, T., Martin, R. (2006) Soft decision combining for dual channel noise reduction. Proc. Interspeech 2006, paper 1059-Wed3FoP.2, doi: 10.21437/Interspeech.2006-555
@inproceedings{gerkmann06_interspeech,
author={Timo Gerkmann and Rainer Martin},
title={{Soft decision combining for dual channel noise reduction}},
year=2006,
booktitle={Proc. Interspeech 2006},
pages={paper 1059-Wed3FoP.2},
doi={10.21437/Interspeech.2006-555}
}