AdaptiveAlgorithms for the Identification of Time-Variant Acoustic Systems

Many digital speech and audio communication systems incorporate models of acoustic systems during signal processing. Often, the time-variant impulse responses of these acoustic systems have to be identified during or before using a communication system by means of adaptive algorithms. This thesis deals with measurement procedures from the field of acoustics and tracking algorithms from communication applications, considers acoustic echo cancellation in hands-free calls or teleconferences, and takes into account additional aspects and difficulties for multi-channel applications compared to single-channel ones. Furthermore, the incorporation of a priori information in the identification process is investigated. Overall, this thesis considers the different aspects of system identification of time-variant acoustic systems for diverse scenarios. It provides new insights into the relationship of different system identification algorithms and proposes novel concepts and algorithms for specific applications.