Abstract
A new active noise cancellation scheme has been developed for the purpose of engine noise cancellation. The major difference between the new approach and the conventional two‐microphone adaptive noise cancellation methods is that the new approach does not require the reference signal. Therefore, the problem of acoustical feedback from the canceling speaker to the reference channel is eliminated. In this new scheme, a linear predictive coding (LPC) technique is employed to synthesize an “engine noise,” which is used as a reference signal, and this synthesized noise is processed by an LMS adaptive filter. Simulation results using a set of digitized engine noise data show that the adaptive filter has a faster convergence rate and more than 10‐dB mean‐squared error improvement by using an LPC noise synthesizer in comparison with using an impulse train as input only. In the case that engine speed is varying or the engine rpm is not available, a multipulse excitation synthesizer has been used to generate the synthesized engine noise. The multipulse excitation method has also provided an improvement of 10 dB in mean‐squared error.
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call