Aspects of source separation in beamforming

One of the most important elements to consider in the application of beamforming is the identification of sources from the beamforming result. This result is often given in form of a map of sound pressure contributions from a number of locations. While on first look the dominating sources may be easily identified, a number of problems arise during practical application: (a) sources closely spaced cannot be separated due to the width of the directional beam, (b) minor sources cannot be detected at all due to the limited signal-to-noise ratio, (c) it is not possible to separate sound pressure contributions from different source mechanisms. Some signal processing methods to deal with these problems are explained in short. The implications of each this methods are discussed. Practical examples using a 32-microphone array are given.