Dopple Group Chat
Dereverberation & Denoise in multi-channel full-duplex communication systems
L. Xin (TU Delft - Electrical Engineering, Mathematics and Computer Science)
R.C. Hendriks – Mentor (TU Delft - Signal Processing Systems)
G. Bologni – Mentor (TU Delft - Signal Processing Systems)
Jaap Haartsen – Mentor
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
In response to the echo and noise interference problems faced by multi-user full-duplex voice communication scenarios, this work proposes a complete solution from signal modeling to beamforming optimization. First, by converting the Bluetooth channel into a virtual acoustic channel, a comprehensive acoustic transmission model cover- ing single-source and multi-source scenarios is established. On this basis, a variety of beamforming and filtering strategies are discussed, and finally, the Multichannel Wiener Filter (MWF) beamforming scheme is proposed. At the same time, for the time-varying spatial domain, the update of the covariance matrix combined with Voice Activity Detector (VAD) is discussed. The simulation shows that the MWF scheme has good robustness and adaptability under different Bluetooth channel latency and low Signal-to-Interference Ratio (SIR) environments; when the user end is equipped with a sufficient number of microphones or there are many bystanders, the system performance is further improved.