PhD Defence by Antonio Andrés Figueroa Duran

Abstract

With the recent advancements in acoustic and audio technology, there has been an increasing demand for thorough characterisation of the sound field properties over large spatial areas, such as entire rooms and auditoria. Sound field control, acoustic heritage preservation, room acoustic design and navigable spatial audio are among many applications that nowadays require knowing the acoustic pressure at a every point in the target domain. However, acoustic data acquisition is an experimentally challenging process, as in order to fully characterise a standard room, the number of required measurements is of the order of hundreds of millions.

This thesis addresses these challenges by developing large-scale sound field reconstruction techniques. These techniques aim to reduce the experimental efforts by predicting the sound field quantities at positions where direct measurements are unavailable. The first part of the thesis focuses on reducing the number of experimental measurements whilst maximising the reconstruction area far beyond the measurement domain. To this end, we propose a method that relies on the generalisable temporal structure of sound in rooms, achieving large-scale reconstruction of individual reflections far from the measurement arrays and preserving the statistical properties of the late reverberation across the entire reconstruction domain.

The second part of the study examines various modelling choices for sound field reconstruction, recognising that the acoustic properties of a domain can vary considerably between rooms. In this context, the thesis delves into the specific nature of the available experimental data to inform and tailor reconstruction approaches. A modular Bayesian formulation is proposed, offeringintuitive control over different physical models, and simultaneously helping to bridge the gap between theoretical acoustics and modern computational methods. The results from this work provide a better understanding of the inclusion of acoustic phenomena in large-scale sound field reconstruction, having an impact in navigable immersive audio, room acoustic design and other areas of acoustics.

Supervisors

• Principal supervisor: Associate Professor Efren Fernandez Grande, Department of Electrical and Photonics Engineering, DTU
• Co-supervisor: Associate Professor Finn T. Agerkvist, Department of Electrical and Photonics Engineering, DTU
• Co-supervisor: Tapio Lokki, Aalto University

External examiners

• Professor Hüseyin Hacihabiboglu, Middle East Technical University, Turkey
• Professor Roland Badeau, Telecom Paris, France

Chair of assessment committee

• Associate Professor Jonas Brunskog, Department of Electrical and Photonics Engineering, DTU

Master of the Ceremony

• TBA