Vergleich robuster Mikrofonarrays

Diploma thesis

The aim of this master thesis is the development of a microphone array to improve speech capture capabilities in conference situations. Therefore, a microphone with several capsules and small aperture will be installed in front of a speaker (e. g. stand alone on the desktop or surface monted on walls). First of all, an overview over basic theorems relating to acoustical array processing and beamforming algorithms is given. Beginning with the simple delay and sum (D&S) approach two more fundamental algorithms, the minimum variance (MV) as well as the filter and sum (F&S) beamformer, are discussed in detail. The advantages and disadvantages of each approach, considering robust beamforming applications, are going to be shown as well.

Using delay and sum principles an array structure is derived that optimizes steering capabilities over a broad frequency range (150 Hz – 8 kHz) under the constraints of small aperture size and minimum number of capsules.

Next, several beamforming theories, shown in recent literature, are illustrated due to mathematical derivation of the algorithms as well as simulations in MATLAB. Furthermore, the simulations shall point out the abilities of the approaches to optimize properties of the beamforming structures concerning to broadband applications. Real time implementations, using Miller Puckettes software Pure Date (PD), are used for acustical measurements on the one hand and informal listening tests on the other. The results will be compared to the theoretical studies. Each approach is evaluated regarding to its robustness against array imperfections. A method to improve the overall robustness will be porposed at the end of this master thesis.

In the last chapter, possibilities to further improve the proposed Robust Adaptive Beamformer (RAB) and prospects to alternative approaches are given.