Causal MSE-Optimal Filters for Personal Audio Subject to Constrained Contrast

Abstract

A novel design method that generates causal precompensation filters is formulated. The resulting filters are constrained with respect to the amount of acoustic contrast they generate and are intended to be used for personal audio. The proposed method provides a more direct method for trading bright zone behavior against acoustic contrast as compared to other causal methods available. It also provides improved control over the temporal properties of the resulting filters as compared to the preexisting non-causal methods. The resulting filters are analyzed by means of simulations, based on measured impulse responses of the sound-system-room interactions. The results of the simulations are compared to simulations of a frequency-domain optimal method with comparable objective, as proposed by Cai et al. and the results of the comparison are explained using the design equations. It is shown that the proposed method is viable, but that unattainable contrasts have a detrimental impact on the spectral bright zone behavior. A few different strategies for dealing with this problem are also proposed. It is demonstrated that the detrimental effect of increasingly strict causality constraints mainly concerns the lower frequency bright zone behavior in the system under investigation, but that the very highest attainable contrast levels may also be reduced somewhat.

Signal-Adaptive and Perceptually Optimized Sound Zones With Variable Span Trade-Off Filters

“…3. Lastly, a control filter length of J = 240 samples (i.e., 15 ms at a sampling frequency of 16 kHz which corresponds to a frequency grid of 66.67 Hz in the ACC method) for all methods, and a segment length of N = 960 samples (i.e., 60 ms at a sampling frequency of 16 kHz) for AP-VAST were used 6 . Therefore, the number of eigenvectors V was in the range from 1 to 1920 since 1 ≤ V ≤ LJ.…”

“…These two zones are created by filtering the signals fed into the loudspeakers, and multiple bright zones can be obtained by superimposing the individual bright and dark zones for every input signal. Many different applications of sound zones have been studied, including outdoor concerts [1], automobile cabins [2]- [6], pedestrian alert systems [7], mobile devices [8], personal computers [9], and other applications [10], [11].…”

“…Since ACC only optimizes the acoustic contrast, it will in general not maintain the spatial characteristics of the desired sound field. Consequently, the ACC methods are most useful in situations where the spatial characteristics are either not important or very hard to reproduce due to complicated, dynamic acoustics environments such as in car cabins [2]- [6]. Various variations of the ACC method have been proposed.…”

“…Various variations of the ACC method have been proposed. These include the energy difference maximization [19], the planarity control [20], subband optimization [3], multiple constraints on the acoustic contrast for different frequency bands [6], and the broadband ACC (BACC) method [21]. The BACC method is different from the other methods in that it operates in the timedomain instead of the frequency-domain.…”

Generation of Personal Sound Zones With Physical Meaningful Constraints and Conjugate Gradient Method

“…T HE creation of personal sound zones became an active 35 area of research ever since it was first introduced approximately two decades ago [1], [2]. Based on this concept, different applications such as cars and aircraft [3]- [9], outdoor concerts [10], mobile devices [11], etc. [12], [13] were studied.…”