New models and approaches for the robust equalization of loudspeaker-room systems

• Applicant: Professor Dr.-Ing. Alfred Mertins
• Subject Area: Acoustics ; Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: from 2015 to 2019
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 262418280

With the growing presence of multiple forms of audio-visual communication, for example, in form of comfortable multi-channel hands-free telephony or by the creation of virtual acoustic environments for multimedia applications, the optimization of loudspeaker-room systems is of great practical importance. The objective of the proposed research is to investigate new approaches for the design of such systems and to enable the equalization of spatial volumes in an optimal way. In previous works of the applicant, it has been successfully shown how acoustic loudspeaker-room systems that suffer from severe reverberation can be effectively equalized by replacing the inverse-filtering paradigm by a shaping of the overall response according to a given decay curve. In ideal cases (e.g. with sufficiently long equalizers) disturbing reverberation can be reduced so far that it becomes inaudible. In the proposed research, the robust design paradigms for listening-room equalization and crosstalk cancellation developed by the applicant are to be extended in such a way that larger movements of listeners can be allowed. For this, novel statistical models for position-dependent system perturbations and compressed-sensing based methods for the estimation (interpolation, extrapolation) of room impulse responses are to be developed and incorporated into room equalization. The goal is, for example, to minimize the number of required measurements while achieving nearly the same robustness as with a complete measuring of the room according to the spatial sampling theorem. For adaptive solutions, the head position is to be captured with an optical tracking system and incorporated into an adaptive equalization. This adaptation founds the basis for the presentation of 3D audio via loudspeakers in reverberant environments, and the proposed model-based approach has the potential to overcome the limitations of existing crosstalk-cancellation systems to nearly anechoic rooms. The limits of performance are to be investigated within the project. To optimally exploit the existing degrees of freedom during filter design, the perception of artifacts will play an important role within the algorithm design.

• DFG Programme: Research Grants