Abstract


This thesis deals with the study and analysis of room acoustics through a process of sound field decomposition sampled with microphones circular arrays.
The acoustic behavior of a room considering all its spatial factors is not straightforward. This problem is known since antiquity: greeks and romans started to look the acoustic qualities in their architectural constructions mainly based on purely practical designs. Later, it was physicist Sabine with his statistical model for measuring the reverberation time of rooms who would lead to the modern analysis of architectural acoustics.
There are currently multiple and advanced analysis methods based on the study of room impulse responses that have contributed to new developments and improvements in the acoustic field. The latest recently proposed methods used microphone arrays to capture more accurately the sound field with all its spatial characteristics.
However they have not yet been adequately exploited to obtain meaningful and relevant conclusions on acoustic rooms and geometric aspects. This thesis conducts an analysis methodology based on plane wave decomposition and modal beamforming techniques. Through the development of a cardioid microphones circular array as well as the implementation of a wave detection algorithm in echograms based on image morphology, the acoustics of a room (reflections, absorptions, etc.) is obtained, analyzed and compared allowing to extrapolate some conclusions about its features, performance and quality. Results obtained by both methods of analysis are compared to identify the most significant reflections of rooms, extracting information about the dispersion and distribution of energy. Furthermore, results are associated with the analyzed room geometry in order to verify the proper work of these methods.
Furthermore, and as an added value in the analysis of room acoustics, a source localization method based on modal analysis is applied under adverse room conditions, i.e. with high coefficient of reverberation and low signal to noise ratio. Results show that the proposed approach is capable of localizing multiple sound sources in very reverberant and noisy environments with high accuracy.
Therefore, this thesis makes a contribution to the study of architectural acoustics presenting a comprehensive strategy that includes both capture and analysis with relevant tools to characterize the most significant reflections that shape their acoustic performance.