Back to list

System parameter estimation of acoustic scenes using first order microphones

Authors Wilding, T.
Year 2016
Thesis Type Master's thesis
Topic Audio Signal Processing
Abstract The aim of this work is to develop a toolbox containing algorithms aiming at performing the self-calibration of a system of distributed synchronized microphone arrays and for examining the geometric features of the surrounding acoustic scene (room inference). The microphone arrays to be used are first order microphones (B-format arrays consisting of four discrete capsules) allowing three-dimensional analysis of the surrounding sound field. The aim of the self-calibration is the retrieval of the microphone positions, with the additional advantage of acquiring the positions of the calibration sources as well. Based on the found microphone and calibration source positions the geometry of the surrounding acoustic scene will be determined, with the examined geometries being either shoe-box shaped rooms or rooms of arbitrary shapes with an arbitrary number of reflective surfaces. The source signals used allow simple repetition while exhibiting an im- pulsive character to ensure easy applicability (claps). By the use of these impulsive source signals, pseudo-room impulse responses can be derived from the microphone recordings without the usually rather time consuming pro- cedures needed for impulse response measurements. From the microphone recordings the directions as well as times of arrival of the direct sound and reflections will be estimated, all of which are then used for the self-calibration and room inference tasks. With the aim of a fast applicability, only real measurements performed in real rooms are used in this work. All computations are performed in Matlab.
Supervisors Schörkhuber, C., Höldrich, R.