This section describes the implementation of the multichannel acoustic beamforming system for meetings based of the filter-and-sum system presented in section 5.1 and also described in Anguera, Wooters and Hernando (2005). This involves the processing of the signal from each available microphone until obtaining the final ``enhanced'' channel output and other related information useful for further processing in the mono-channel speaker diarization system presented in the next section.
Figure 5.3 shows the different blocks involved in the filter-and-sum process. The system is able to handle from 2 microphones to as many microphones as memory allows in the computer where it is executed. Each processing stage is either performed on each individual microphone or to all microphones in combination. Each processing block is described in detail below.