Psystem design and optimization for in-car environments

Abstract

The usage of loudspeaker arrays to provide individual audio experiences in shared envi-ronments represents an innovative approach to deliver personal messages or music to a single listener without the need of wearable and uncomfortable devices. In the last decades, the incessant pursuit of solutions to tackle this problem has led to the development and the continuous improvement of sound field control methods, such as the Acoustic Contrast Control (ACC) and the Pressure Matching (PM). Although these novel methods perfectly work in ideal scenarios, they present strong limitations in real and reflective environments. The main problem is related to the poor sound field control robustness with respect to en-vironmental changes and listeners movements from the sweet spots. In fact, the presence of reverberant tails in the transfer functions between loudspeakers and microphones leads to long and resonating sound field control filters, which lead to a low quality sound field control. In this thesis the sound field control problem for a personal audio system design and development is treated and analysed for in-car environments, being an interesting solution to provide personal sound experiences to all the passengers independently. The sound field control problem is split in two parts, namely the sound zone separation and the binaural reproduction problem. For both the cases the performances of the proposed sound systems are analysed in the ideal scenario, namely the free-field. Then, once the upper limit of our system has been defined, the sound field control performances in anechoic conditions are compared with the ones obtained inside the car, in order to define the main losses introduced by the presence of a variable sound field caused by all the reflections. The purpose of this thesis is to analyse and optimize the sound field control capabilities of a loudspeaker array system installed inside the car cockpit. For this goal a technique to optimize the PM sound field control through the measured transfer function pre-processing is presented. This method is based on a frequency-dependent trimming of the impulse responses measured and used to compute the PM control filters. The idea is to neglect from the computation those transfer function components that have a great variability in the space within the test zone, namely the late reflections that are present especially in the cockpit environment.