Q. How does virtual surround work?

Published in SOS July 2004
Printer-friendly version

Sound Advice

Ivo Eames

Technical Editor Hugh Robjohns replies: The more effective 3D-from-stereo systems, such as those from Q Sound Labs and those incorporating Roland's RSS technology, make use of complex comb filtering of the signals in the two channels, with the idea of simulating the natural comb filtering that occurs as sound enters the ears.

Q Sound's QSys/AX Direct X plug-in produces 3D-from-stereo effects. Q Sound's QSys/AX Direct X plug-in produces 3D-from-stereo effects.

The term 'comb filter' refers to the characteristic shape of the frequency response which is created when a signal is combined with a slightly delayed version of itself — if you were to plot the frequency-response graph of the combined signals, you would see regularly spaced deep and narrow notches that look a little like the teeth of a comb. The resulting effect is often described as sounding 'phasey' or like it's coming down a tube.

Our hearing system primarily makes use of level and time-of-arrival differences between sounds arriving at the listener's two ears to help locate the direction of sound sources, but there is also a third technique that relies on comb filtering which is, as yet, still not completely understood.

Essentially, as sound enters the ears it also reflects off the curves of the pinnae (the curved cartilage that forms the outer ears) and possibly also the shoulders, and those reflections combine with the direct signal, create very subtle but unique interference comb filtering.

From a very early age our brains gradually learn how to relate specific patterns of comb-filter notches to the corresponding sound source directions, and it can be a remarkably accurate system. However, everyone's ears are shaped differently and so everyone has slightly different associations between comb-filter pattern and corresponding source direction. This makes it virtually impossible for any system generating artificial comb-filtering artefacts — such as RSS and Q Sound — to produce reliable and repeatable 3D positional impressions for more than one person.

As a result, although both Q Sound and RSS systems can create the impression of sounds located in 3D space, different people often perceive the same sounds as coming from entirely different places — and sometimes with wildly differing directions. Sound reflections from room boundaries and mixing console surfaces, for example, can also affect the spatial imaging quite dramatically — the deader the listening environment, the more stable and believable the effect will be.

So, if you just want a spatial effect these kinds of systems can work quite well. However, if you want accurate 3D imaging from two speakers that will be heard the same way by lots of different people listening in different environments, current systems are probably not the way to go — binaural techniques via headphones are probably a better solution.