Originally Posted by AndreYew
And the proper signals to drive them. The bass must be decorrelated to 40 Hz. Most commercially available recordings have mono (ie. perfect correlation) bass, unfortunately.
This is indeed true. With monoed bass you achieve nothing meaningful.
Bass in any large listening room is, however, nothing near correlated as you expect, per measurements made and published, among other places, in Kuo, S. and Johnston, J. D., A study of why cross channel prediction is not applicable to perceptual audio coding, 111 AES Convention, New York, NY, 21-24 September 2001.
While that is not definitive, it shows the degree of decorrelation using part-velocity, part-pressure mikes in a near coincident array, even into bass frequencies. This shows that what hits your ear in the ORIGINAL venue is not correlated. THAT, coupled with the out-of-phase demonstration, is pretty conclusive, really.
Now, artificial stuff, say using a hilbert on bass (an interesting proposition, that, too
), may not do what you want or expect. What you want is a time characteristic much like some set of points in the original venue, not mono bass, not mono bass that's been processed.
The argument "only under special circumstances, etc" is simply not true.
We had a very nice demo at AT&T before it shut most of itself down, in which 1 and 2 channel renderings of a playback (taken from an array mike) were very, very boomy. Note, even 2 channel with 2 woofers was boomy. The 5-channel rendition, on the other hand, puts the room to the sides and back, and the boom hangs in the back like one expects, and sounds like a good concert hall rather than boomity-boom. This was with a set of 1 to 10 full range (Snell C-v) loudspeakers.
There was no sub in this setup at all.
As for the boominess, in fact the listener was getting the same average spectrum at each ear in both 2 and 5 channel, to a very notable degree. (there were some minor cancellations in mono in the bass)