Noise Robust Speech Coding

Digital Communications The presence of acoustic background noise can substantially degrade the performance of a digital communication system. This is especially true when a model-based speech coder is used in the communication link. Such coders are tuned to specific characteristics of clean speech signals and they handle noise corrupted speech signals rather poorly. Currently, we study novel noise reduction methods based on rank-reduction techniques [M. Jeppesen, C.A. Rødbro, S.H. Jensen, 2001 (b)] and iterative Wiener-filtering techniques as well as the integration of these techniques into different speech coding algorithms. The long-term research goal is to design low-rate and variable-rate noise robust speech coders that can be applied in wireless IP-based communication systems. (Søren Holdt Jensen)