This paper presents a new multi-channel dereverberation system combining a modified correlation-based blind deconvolution with the multi-microphone spectral subtraction. In the proposed system, we make M combinations of observed signals and apply them to the correlation-based blind deconvolution. The deconvolved signals are then used as inputs to the multi-microphone spectral subtraction. The proposed system estimates the reverberant energy by using both a frame delay and a frequency-dependent weight. Due to accurate estimation of the reverberant energy, the combination of the correlation-based blind deconvolution with the multi-microphone spectral subtraction provides improved dereverberation performance. Performance improvement of the proposed system has been confirmed through experiments.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.