The application of adaptive filters to active sound control

Active sound control involves the introduction of a number of secondary acoustic sources whose outputs are arranged so as to destructively interfere with the sound from some `primary´ source. The technique works best with low frequency periodic signals where the wavelength is long and the waveform predictable. The method used is to adapt the output from a number of loudspeakers so as to minimise the sum of the squares of the pressures at a number of microphones. This may be considered as an adaptive filter problem with a pure tone reference signal being fed to the loudspeakers via a bank of digital filters. The paper begins with a review of a multichannel LMS algorithm suitable for this task and goes on to explain some of its characteristics including results measured with a practical system feeding 16 loudspeakers to control 32 microphones