A customized beamformer system for acquisition of speech signals

The authors describe algorithms and hardware used to implement an acoustic beamformer for the purpose of speech acquisition. Beamforming algorithms studied include conventional delay and sum, Frost, and the generalized sidelobe canceler. In each case, parameters related to the algorithm such as the number of sensors, the number of tops, and the rate of adaptation were studied in order to optimize performance with speech signals yet maintain feasibility of real-time implementation on a single digital signal processing (DSP) chip. Results with simulated and real data as well as with an audio tape are presented. Eight channels of microphone data were simultaneously digitized using TLC32044 analog-digital converters, multiplexed in a T1 data frame format, and sent to a TMS320C30 serial port for subsequent processing