Low-Complexity Neural Speech Dereverberation with Adaptive Target Control
Nagashree K. S. Rao, Srikanth Raj Chetupalli, Shrishti Saha Shetu, Emanuël A. P. Habets, and Oliver Thiergart
International Conference on Acoustics, Speech, & Signal Processing (ICASSP) 2025.
Abstract
Existing neural network-based speech dereverberation approaches use a fixed-length early reflection part of the reverberant signal as the target for estimation, irrespective of the severity of reverberation. Such a strategy often leads to distortions in the enhanced signals in highly reverberant scenarios. In practice, while some listeners prefer minimal speech distortions, others have a higher tolerance for distortions and prefer a clean signal. To address these points, we propose a novel target definition and a low-complexity neural network for user-controlled single-channel dereverberation. Our target definition is parameterized by the relative amount of reduction in the late reverberation energy. Further, the same parameter is passed as a control input to the dereverberation network for adaptability during inference. Objective and subjective evaluation shows the feasibility of the proposed dereverberation approach.
For more information about this technology please contact Prof. Dr. Emanuël Habets (emanuel.habets@audiolabs-erlangen.de).
Audio Examples
** Example 1
** Example 2
** Example 3
** Example 4
** Example 5
** Example 6
** Example 7 (Noisy reverberant speech example)