Search

Fan noise in acoustic duct shows narrowband dominated property, the low frequency periodic part of fan noise is usually unstable and changing, which may lead to frequency mismatch for the narrowband active noise control of fan noise. In this paper, a narrowband active noise control system with a frequency compensation is developed. The noise signal on several positions of the duct are collected to select suitable positions for microphone setting and to assess the frequency mismatch in the system. A momentum LMS is introduced for a faster tracking and compensating of the frequency at the cost of little computation burden. Extensive experiments are conducted to demonstrate the superior performance of the proposed system as compared with the traditional frequency compensation based system.

Wearable electronic systems have been and will continue to be utilized in both civil and military uses. Strong noise environments derived from large vehicles (e.g. ships, aircrafts, or military tanks) seriously affect the quality of speech communications especially for wearable systems without hermetic packaging. Bone conduction technology through the acquisition of skull vibration is able to obtain voice information, which can effectively avoid the interference of acoustic noise on the speech. In this paper, a high-performance wearable bone-conducted speech enhancement system is developed to reduce the distortion of the environment noises. Both bone-conducted and air-conducted voices are used to train the equalization function of bone-conducted speech to air-conducted speech based on the deep neural network, and spectrum coefficients of linear predictive coding is taken as feature information for conversion model.

The feedback Active Noise Control (ANC) is mainly used to reduce tonal noise, but for broadband noise like chaotic noise, it is less involved other than using filtered-s least mean square (FSLMS) algorithm. Though the feedback ANC based on FSLMS algorithm is successfully used to reduce chaotic noise, there are still some drawbacks such as slower convergence speed, larger steady-state error and heavier computational load. To solving these problems, a novel feedback ANC system using recursive FSLMS (RFSLMS) algorithm is proposed to reduce chaotic noise in this paper. Compared with conventional feedback ANC based FSLMS algorithm, the proposed scheme holds faster convergence speed and less steady-state error. Extensive simulations are conducted to demonstrate the effectiveness of the proposed nonlinear feedback ANC based RFSLMS algorithm.

Sound source location based on General Cross-correlation can be divided into two parts—time delay estimation and location estimation. Considering the cost and calculation speed, the location system based on GCC receives more and more concern from researchers. This paper provides a new approach to improve the accuracy of location estimation using signals from two microphones. For that reverberation carries information of location of acoustic source, we can separate the detail which contains information of sound source under assumption of knowing coordinate of microphones. This method can be applied to many aspects especially for humanoid robot.

By analyzing the theory of functional link artificial neural network (FLANN) structure based on filtered-s least mean square (FSLMS) algorithm which is usually used in the nonlinear active noise control (NANC) system, it can be found that the controller coefficients of nonlinear parts are multiple related, this problem causes much unbalance to calculate these coefficients and restraints the performance of FSLMS algorithm. To solve this issue, a modified FSLMS (MFSLMS) algorithm is proposed in this paper, it can weaken these multiple relationships greatly by adding a corrective filter before trigonometric expansion. Compared with conventional FSLMS algorithm and its other improved versions, MFSLMS algorithm not only performs better on noise cancellation, but also has less computational complexity. Extensive simulations are conducted to demonstrate the effectiveness of the proposed algorithm.

In this paper, a residual error separation (RES) scheme using variable leakage LMS algorithm is proposed for a complete parallel narrowband active noise control (ANC, NANC) system. The RES subsystem is dedicated to separating individual sinusoid being targeted from the residual noise. Each separated sinusoid, instead of the whole residual noise, is used to update the corresponding control filter weights, which may significantly improve the overall performance of the system. A variable leakage (VL) LMS (VL-LMS) algorithm is developed for the RES subsystem to maintain its stability and to improve the system performance. Extensive simulations are provided to confirm the superiority of the proposed NANC system.

Narrowband active noise control (ANC) systems enjoy good performance where sinusoidal signals dominate in the primary noise, on condition that a reference signal of the same frequencies with the primary noise is given. However, frequencies of the reference signal provided by nonacoustic sensors are usually different from that of the primary noise due to temperature changes, aging, etc. Such frequency mismatch (FM) will make the narrowband ANC systems unable to suppress the primary noise effectively, even render them useless. In this paper, we propose a new narrowband ANC system that integrated with a frequency estimation subsystem. The frequency estimation is obtained from a spectrum computation based on an adaptive linear prediction filter. The estimated frequencies are used by the cosine signal generator to produce a more accurate reference signal to the main controller, thus the performance deterioration caused by FM can be mitigated. The effectiveness of the proposed system has been confirmed by numerous simulations.

Bone conduction has been used for speech enhancement in very noisy environments. It usually includes nonlinearity in transmission, small non-stationary noise due to body movements (friction, collision, wind noise, crosstalk and so on), serious attenuation of high frequency components, etc. These problems usually lead to poor intelligibility of bone-conducted (BC) speech. Recoverying an air-conducted (AC) speech from a bone-conducted recording alone has been considered to be a very difficult task. In this paper, we propose a nonlinear adaptive noise canceller (ANC) that uses both bone- and air-conducted measurements for speech recovery. In this ANC, bone-conducted measurement is used as the reference signal while the air-conducted one is adopted as the primary signal, and a functional link artificial neural network (FLANN) is introduced as the nonlinear adaptive filter. Application to real speech signals is conducted to confirm the effectiveness of the proposed system.

The conventional broadband active noise control (ANC) system may not be able to function well when the target noise contains both wideband and narrowband components simultaneously. Feed-forward hybrid active control has attracted quite a lot attention recently, as it can effectively suppress both components at the same time. Such hybrid ANC system consists of three tightly integrated subsystems, namely, the sinusoidal noise canceller (SNC), broadband ANC (BANC) system and narrowband ANC (NANC). The BANC and NANC subsystems are placed in a parallel form, such that a good performance can be achieved. What's more, the update formulas of the two subsystems depend on the overall residual noise of the whole system, which is a combined measurement of two individual errors that correspond to the wideband and narrowband noise reduction, respectively. Using this "global" residual error may make both subsystems unable to reach their best performance. In this paper, a new hybrid ANC system is proposed to tackle this problem, by adding a new residual error separation (RES) subsystem to the conventional hybrid ANC system such that the two subsystems can update themselves based on their own residual errors. In this way, the two subsystems can work more effectively and make the whole system more capable. Extensive simulations show that the new system enjoys improved performance.

Adaptive Fourier analysis has numerous applications in biomedical engineering, music signal processing, digital communications, power engineering etc. So far, a lot of adaptive algorithms and systems have been developed and applied. In this paper, a variable step-size LMS (VSS-LMS) algorithm is proposed for adaptive Fourier analysis of noisy sinusoidal signals. It significantly outperforms the conventional LMS and p-power algorithms in both stationary and nonstationary environments at the expense of very little increase in computational cost. Extensive simulations as well as application to real noise signals generated by large-scale factory rotating machines are conducted to confirm the improved performance and tracking capabilities of the proposed algorithm.

The well-known broadband active noise control (BANC) system using an FIR filter usually works well when the primary and reference noise signals are periodic in nature. This paper investigates the mean convergence of weight errors of the BANC system in the presence of multiple harmonic excitation. Difference equations governing the mean convergence of the system are derived and discussed in detail. The obtained analytical results enrich our understanding of the BANC operated in an environment with multiple sinusoids. Extensive simulations are conducted to support the analytical findings.