| Tartalmi leírás: |
Microphone Array Source Separation is a wellknown topic in the field of Acoustic Signal Processing. In the literature, there are two main approaches to this problem: Blind Source Separation (BSS), and Acoustic Beamforming. Beamforming methods are specifically designed for array signal processing, they achieve source separation by altering the directionality of the microphone array. In contrast to this, BSS methods such as Independent Component Analysis (ICA) only use the statistical independence of the sources without considering the properties of the microphone array (significantly higher number microphones than sources and known geometric structure). This paper investigates the possibilities of optimizing ICA based source separation for microphone array usage. We propose novel solutions to improve the separation performance of ICA based BSS using the array properties, and evaluate the performance of the proposed method on real microphone array recordings.
|
