Controllo attivo di zone sonore personali mediante elaborazione digitale del segnale

The advancement of audio technology in a world characterized by increasing connectivity and constant background noise has led to the development of personal sound zone systems. These systems are designed to provide tailored audio experiences to individual users in shared spaces. By utilizing several loudspeakers, personal sound zone systems can deliver distinct audio signals to various zones without the need for headphones, improving listener comfort and privacy, and reducing noise pollution. Personal sound zone systems find applications in various scenarios, including in-vehicle audio, public and home entertainment, and communication. They have the potential to offer benefits, such as individualized sound zones for passengers during car or public vehicle journeys, separate audio areas for multiple individuals at home and public spaces, and enhanced audio quality for phone calls, video conferencing, and online meetings. However, designing and implementing personal sound zone systems presents several challenges, including achieving high acoustic contrast among zones, minimizing signal distortion and interference, optimizing array configurations, and adapting to changing user preferences. To control the sound field in personal sound zone systems, advanced digital signal processing techniques are required. These techniques have applications in immersive audio experiences, virtual and augmented reality, gaming, noise cancellation, and personal sound zone generation. This thesis deals with various aspects of a personal sound zone systems with application in the automotive scenario. Related to the capabilities of personal sound zone systems to control the soundfield in a particularly reverberant scenario, such as the cabin of a vehicle, this research introduces four methods for processing the measured impulse responses, reducing late reflections to enhanceacoustic contrats, robustness, and sound quality. Various methods have been proposed for sound field control, such as pressure matching and acoustic contrast control techniques. For a part of this work, the personal sound zone method is adopted, however, also the performance of the acoustic contrast control method in terms of sound quality in a real-world system is investigated. Moreover, two techniques derived from the pressure matching method are proposed to improve the acoustic contrast, reproduction error and robustness in the considered scenario. With the purpose of achieving a high fidelity of the reproduced audio, an acoustic pressure with flat spectrum and constant group delay is considered as target in most of the literature related to pressure matching. However, this may not be the optimal target in order to improve acoustic contrast maintaining high fidelity reproduction. For this reason, the first proposed technique involves the optimization of the target phase for pressure matching with the aim of acoustic contrast maximization. One of the disadvantages of the original formulation of pressure matching is that the performance is not robust against errors in the measurement positions with respect to the realistic positions of the listeners. With the purpose of solving this weakness, a statistical pressure matching algorithm is developed. This technique allows to improve the reproduction fidelity and acoustic contrast by using several measurements to average out the effect of the errors. A significant aspect of personal sound zone systems involves designing filters to control the audio signals at the inputs of the loudspeakers based on the acoustic responses and assessing the resulting performance. Early evaluations were often performed under ideal conditions, thus, overestimating the system achievable performance in realistic conditions. To address these limitations, a stochastic model is proposed to generate mismatched frequency response for realistic performance prediction. This model considers complex coefficients in the frequency domain and perturbs the acoustic responses in the performance evaluation step.