WFS differs from other spatialization techniques as it can theoretically reproduce the wavefront curvature associated with a virtual source in front of, or behind the loudspeaker array. As WFS recreates the sound field over an extended area, the amplitude and location of a virtual source should therefore change in a realistic fashion with listener movement. It has been suggested that this idea of motion parallax, where the perspective changes naturally as the listener moves, can provide an indication of the distance of a WFS virtual source.
Boone assessed the intelligibility of speech by comparing noise and speech signals played back using a single loudspeaker, and two WFS virtual sources which differed only in terms of their distance [Boone et al, 2003]. The results showed an improvement in intelligibility when the speech and noise source were separated using WFS which would seem to indicate that there is at least some perceptual difference between two WFS virtual sources differing only in distance. However, Wittek points out that this perceptual difference could occur due to differences in the way a single loudspeaker and an array of loudspeakers interact with the acoustic of the
reproduction room [Wittek, 2003].
Nogues conducted an experiment which provided better evidence that the movement of the listener through a WFS sound field does indeed provide some indication of distance [Noguès et al, 2003]. They asked the subjects to control the distance of a WFS virtual source so that the source distance matched the distance of
two other simultaneously reproduced virtual reference sources (see Figure. 6.24). The subjects were asked to move around in the listening area throughout each test. In the first experiment, the subjects could only manipulate the WFS source distance while other parameters such as the direct to reverberant energy ratio and signal level were kept constant. The results showed that the subjects were indeed able to position the middle guitar in between the other two solely using the perspective cue of the WFS sound field. This result suggests that the perception of distance due to the virtual source position in a WFS system can be perceived independently of the subjective distance impression.
Fig. 6.24 Technical setup for listening tests by Nogues
In a second experiment, other distance cues such as the direct to reverberant ratio were included. A number of sources were synthesized at different distances and the subjects were asked to adjust the direct to reverberant ratio of each source so that the perceived distance matched that of a pair of reference sources. The results showed that the perception of distance was primarily influenced by the direct to reverberant ratio, and that the wavefront curvature is a weak localization cue which is easily overridden by other cues.
Kerber implemented listening tests to compare the perceived distance of real and virtual focussed dry sources. The results for a real source shown in Figure 6.25 support the results of other tests in that only small distances ( < 1m) are perceived accurately and that large distances are consistently underestimated. Focussed WFS sources do not seem to be able to produce the same distance perception as real
sources. Wittek suggests that these results indicate that at a fixed listening position, the curvature of the wavefront of a dry WFS virtual source does not support distance perception. However, in spite of not being a crucial cue, a correct wavefront
curvature (and thus a consistency between curvature and actual distance) may support the perception of distance, particularly if the listener can move [Wittek, 2003]
Fig. 6.25 Distance of real (left) and virtual (right) sources reported by Kerber
Usher similarly found that in the absence of any indirect sound, when a source is positioned beyond a certain distance using a WFS system, the curvature of the wavefront does not seem to be used to determine the distance of the virtual source, but rather the timbre of the perceived source dominates [Usher et al, 2004]. It would seem, therefore, that in order to accurately produce WFS virtual sources at different distances, some form of artificial reverberation is needed to provide additional distance cues. However, Wittek points out that disturbing reflections caused by the WFS array itself may in fact also hinder the perception of the distance of virtual sources in front of or behind the array. It is not the case that a dry WFS virtual source will automatically produce a natural reflection pattern in the reproduction room [Wittek, 2003] and this is illustrated in Figure 6.26 which shows a WFS system with a virtual source (blue dot) positioned in front of the array. The correct reflections that would arise if a real source was at this position are indicated by the green dots, while the actual reflections that arise are shown as orange dots. Clearly both the timing and direction of the actual reflections do not correspond to the desired source position, but rather to the distance of the loudspeaker array itself. The virtual source distance will
than the specified source distance. This again indicates the significant influence of the reproduction room acoustic on the performance of WFS systems. Various listening room compensation schemes have been proposed which can actively cancel early reflections in the horizontal plane [Corteel et al, 2003; Spors et al, 2003]. The results of simulations suggest that these techniques could help to reduce the
detrimental effect of early reflections in the listening room over a large area.
Fig. 6.26 WFS reproduction room reflections
Or course, WFS can also simulate different source distances through the addition of early reflections and reverberation. Figure 6.27 illustrates a scheme proposed by Caulkins to artificially inject early reflections which are otherwise absent in the WFS reproduction of focussed sources [Caulkins et al, 2003]. As with other spatialization techniques, more listening tests are required to fully determine the perceptual effect of combined and potentially conflicting virtual and real acoustic reflections.