It is often difficult to position microphones at the ear position of listeners to directly monitor the perceived sound, in local active sound control systems for example. The pressure at these positions can be estimated using virtual sensing with an array of remote monitoring microphones, however, if some assumptions are made about the sound field. In active control, the sound field due to the secondary sources can be reasonably easily accounted for but the primary sound field, which is to be controlled, will in general be due to a number of potentially correlated primary sources, whose positions are unknown and may vary in time. The virtual sensing in this application thus needs to be robust to the properties of the primary sound field, both in the choice of the remote monitoring microphone positions and in the design of the filters used to process these to estimate the pressure at the desired position. If the controller is feedforward, the causality of these filters may also be relaxed if the adaptive algorithm is designed to minimise a delayed virtual error signal. This paper describes examples of such robust design, particularly applied to the local active control of road noise in vehicles.

Personal Audio systems can be used to provide information and entertainment content in public spaces. Limitations in array directivity mean that speech information intended for a target region may remain intelligible elsewhere. This compromises privacy for target listeners and could prove distracting or annoying to passive listeners nearby. A system has previously been proposed whereby the intelligibility of this leaked speech is reduced by radiating an artificial masking signal into the dark zone; this masking signal has been optimised to minimise the potential for annoyance whilst achieving a predefined level of intelligibility in each zone, but only free-field responses were considered. In practice, systems located in public spaces will be adversely affected by noise and reverberation. This detriment to system performance can be quantified using engineering measures such as acoustic contrast, although the perceived performance as evaluated by users does not necessarily correspond. The present paper explores the effect of using analytical and measured transfer responses on speech intelligibility and system optimisation using a practical example of a personal audio system in a room.

The broadband control of noise and vibration using multi-input, multi-output (MIMO) active control systems has a potentially wide variety of applications. However, the performance of MIMO systems is often limited in practice by high computational demand and slow convergence speeds. In the somewhat simpler context of single-input, single-output broadband control, these problems have been overcome through a variety of methods including subband adaptive filtering. This paper presents an extension of the subband adaptive filtering technique to the MIMO active control problem and presents a comprehensive study of both the computational requirements and control performance. The implementation of the MIMO filtered-x LMS algorithm using subband adaptive filtering is described and the details of two specific implementations are presented. The computational demands of the two MIMO subband active control algorithms are then compared to that of the standard full-band algorithm. This comparison shows that as the number of subbands employed in the subband algorithms is increased, the computational demand is significantly reduced compared to the full-band implementation provided that a restructured analysis filter-bank is employed. An analysis of the convergence of the MIMO subband adaptive algorithm is then presented and this demonstrates that although the convergence of the control filter coefficients is dependent on the eigenvalue spread of the subband Hessian matrix, which reduces as the number of subbands is increased, the convergence of the cost function is limited for large numbers of subbands due to the simultaneous increase in the weight stacking distortion. The performance of the two MIMO subband algorithms and the standard full-band algorithm has then been assessed through a series of time-domain simulations of a practical active control system and it has been shown that the subband algorithms are able to achieve a significant increase in the convergence speed compared to the full-band implementation.

Active sound profiling, or active noise equalization strategies have been proposed to achieve spectral shaping of a primary disturbance signal. The control algorithms proposed to achieve such spectral shaping have either suffered from poor robustness to plant modeling uncertainties or required high levels of control effort. To improve the robustness of active sound profiling to uncertainties in the plant model, while avoiding increased control effort, a modified phase-scheduled-command filtered-x least-mean-square algorithm is proposed in this paper. The new algorithm provides improved stability, while requiring the minimum control effort. This improvement is achieved by replacing the plant model with an intelligent adaptive-hysteresis switching mechanism to allow the necessary estimation of the disturbance signal phase. The improved performance and robustness of the proposed algorithm is demonstrated through a series of simulations using measured acoustic responses.

This paper describes a time domain design method for calculating the coefficients of FIR filters used to drive a loudspeaker array for personal audio. A motivating application is to boost the television audio in a certain spatial region with the aim of increasing the speech intelligibility of the hearing impaired. As the array is of small size, superdirective beamforming is applied to increase the low and mid frequency directional performance of the radiator. The filters for such arrays have previously been designed one frequency at a time, leading to non-causal filters that require a modeling delay for real time implementation. By posing the filter optimization in the time domain, the filter responses can be causally constrained, and the optimization is performed once for all frequencies. This paper considers the performance of such filters by carrying out off-line simulations, firstly using the impulse responses of point sources in the free field, and then with the measured responses of a loudspeaker array in an anechoic chamber. The simulation results show how the time domain optimization allows the creation of filters with either a low order or a low modeling delay.

In the above paper Chang and Kuo show that the convergence of parallel narrowband active noise control systems is limited by the interaction between the multiple narrowband adaptive filters. To overcome this problem they propose a new algorithm that provides separate error signals to each of the narrowband adaptive filters, generated by filtering the error signal using bandpass filters tuned to the control frequency of each of the adaptive filters. Although these bandpass filters do not introduce an additional phase shift at the control frequency they do introduce an additional group delay into the secondary path. It is shown in this correspondence that although introducing these bandpass filters allows an increase in the step size it may also limit the convergence speed compared to the conventional algorithm.

As well as being able to reproduce sound in one region of space, it would be useful to reduce the level of reproduced sound in other spatial regions, with a “personal audio” system. For mobile devices this is motivated by issues of privacy for the user and the need to reduce annoyance for other people nearby. Such personal audio systems can be realized with arrays of loudspeakers that become superdirectional at low frequencies, when the array dimensions are small compared with the acoustic wavelength. The design of the array then becomes a compromise between performance and array effort, defined as the sum of mean squared driving signals. Various methods of formulating this tradeoff as a regularization problem have been suggested and the connection between these formulations is discussed. Large array efforts are due to strongly self-cancelling multipole arrays. A concern is then the robustness of such an array to variations in the acoustic environment and driver sensitivity and position. The design of an array that is robust to these uncertainties then leads to a generalization of regularization.