Publication | Closed Access
Active adaptive sound control in a duct: A computer simulation
381
Citations
0
References
1981
Year
AeroacousticsAdaptive FilterEngineeringNoise ControlAerospace EngineeringAudio Signal ProcessingMechanical SystemsNoiseSystems EngineeringSpeech EnhancementSpeech ProcessingControl System ElementsNoise ReductionSound PropagationSound CancellationActive Noise ControlSignal ProcessingComputer Simulation
Active sound cancellation systems typically rely on open‑loop control, near‑zero phase delay, and a fixed acoustic transit time between microphone and loudspeaker. This study explores the use of closed‑loop control in a duct‑based sound cancellation system via digital simulation. The method extends Sondhi’s adaptive echo canceler and Widrow’s adaptive noise canceler to control, employing an LMS gradient‑search algorithm. Simulation results show that the closed‑loop algorithm can cancel pure tones to the computer’s bit‑noise floor (−120 dB) and, even with additive white noise, suppress tones at least 10 dB below the noise level for SNRs down to 0 dB; the theory also predicts effective tracking of slowly varying tones and cancellation of narrow‑band sound amid overlapping broadband noise, suggesting broader applicability to systems with transport delay.
Most active sound cancellation systems reported in the literature use open-loop control, depend on near-zero phase delay in control system elements, and require constant acoustic signal transit time from a signal pickup (microphone) to a control sound source (loudspeaker). The applicability of such systems can be significantly enhanced by using closed-loop control. This study concerns a digital computer simulation of adaptive closed-loop control for a specific application, sound cancellation in a duct. The key element is an extension of Sondhi’s adaptive echo canceler and Widrow’s adaptive noise canceler from signal processing to control. The adaptive algorithm is thus based on the LMS gradient search method. The simulation shows that one or more pure tones can be canceled down to the computer bit noise level (−120 dB). In the presence of additive white noise, pure tones can be canceled to at least 10 dB below the noise spectrum level for SNR’s down to at least 0 dB. The underlying theory implies that the algorithm allows tracking tones with amplitudes and frequencies that change more slowly with time than the adaptive filter adaptation rate. The theory implies also that the method can cancel narrow-band sound in the presence of spectrally overlapping broadband sound. The method can be applied more widely, particularly to control systems that involve transport delay.