Resumen
Office employees are exposed to acoustic noise, especially in an open office environment. Passive or active headsets sufficiently reduce the level of noise, but their long-time wear is uncomfortable. The paper presents an active headrest system built into a chair. Feedforward control is utilized with multiple reference microphones, and flexible gooseneck microphones serve as error sensors. The reference sensors surround the chair, allowing the suppression of sound waves arriving from any direction. The concept of multiple reference control has been tested by extensive experiments showing that multiple reference signals help to increase the suppression in normal rooms where reverberation occurs, even if only one noise source is present. The preliminary experiments are completed by a series of simulations aiming to explore the zones of quiet at the user’s ears. The paper introduces the construction details of the noise-canceling chair: The two loudspeakers are controlled by the signals of two error microphones and four reference signals. The controller is based on the normalized filtered error least mean squared algorithm, implemented on an Analog Devices ADSP-21262 signal processor-based hardware. Experimental results are reported that show the efficient suppression of tonal, as well as broadband disturbances.