rnhood,
But, I do know that the better speaker manufactures like Thiel strive to get that group delay figure low as possible. There must be something to it.
There is an excellent AES paper that goes a long way in explaining the low frequency group-delay issue. It is:
Preprint Number: 2056 Convention: 74 (September 1983)
Author: Fincham, L. R. If you're really interested you can purchase it from AES for $20. Other than that I will attempt to give the short version here.
Basically, Fincham of KEF fame built both a digital recording system and loudspeaker playback system that was flat down to a few Hz.(approx. 5Hz) Obviously that took a lot of big woofers in a big room, but they did it.
It is a known engineering fact that if one extends bandwidth down to 0Hz (or very close) the group-delay will be essentially "zero." That means linear phase and zero signal delay through the system at all frequencies.
Then they recorded an orchestra (I believe) and played it back through the speaker system. They had a number of volunteers give their impression of the sound. The most noted aspect was that virtually everyone commented on the apparent "lack" of bass!
Then they inserted a very low distortion electronic filter that was designed to intentionally inject adjustable amounts of low frequency group-delay/excess phase - while keeping the magnitude response flat. As the amount of delay increased listeners reported a subjective increase in the amount of bass.
The upshot of this test confirmed that the ear/brain mechanism interprets increasing amounts of low frequency delay as "more bass." Albeit "artificial bass" - we transform the difference in delay of low frequencies with respect to high frequencies into "magnitude." A simplified equation would be: T(lf) - T(hf) = Magnitude. As low frequencies get "stretched out" or disconnected in time from the higher frequencies, our brains decide there is a greater amount of low frequency information.
And all of this makes sense. There is a preponderance of data that suggests human hearing is time dependant as well as frequency dependant. The real issue is a matter of degree. Some would suggest that very small delays (less than 1mS) are clearly audible and use this to promote the idea that only 1st order - 6dB/oct. crossovers should be used in order to avoid them. The scientific data does not support this claim though - or if at all, its importance is extremely small.
On the other hand, signal delays at low frequencies can be on the order of ten's of milliseconds. It is not uncommon to see delays of 30mS or more in bass-reflex systems tuned to 40Hz - give or take. It only makes sense that such large delays would be audible.
Many companies take advantage of this little psycho-acoustic phenomenon - intentionally or not. "More bass" sells more speakers. On the other hand, we here at SP Tech know that such "bass" is not accurate. In fact, it is one major source of that "bloated," 'boomy," "one note" bass effect.
In order to minimize it in our systems, we tune our reflex enclosures as low as possible. As in Fincham's experiment above, given a certain type of alignment as a constant (reflex, sealed, etc.), when one extends bandwidth to lower frequencies, one reduces the total amount of low frequency signal delay as well.
This is not true though when mixing alignments. Even though our Continuum 2.5 has the same -3dB point as our Revelation (25Hz), the rolloff "slope" (24dB/oct.) is twice that of the Revelation (12dB/oct.) - due to the fact that it is a reflex design in contrast to the Revelation's transmission-line. Consequently, since the "F3s" are the same, the Revelation will have exactly 1/2 the amount of group-delay as that of the Continuum 2.5.
Is the difference audible and worth the extra expense of the Revelation? Audible? - yes, to the experienced listener that knows what "live" bass sounds like. Worth the price of admission? - Well...as far as we're concerened, we wouldn't offer it if we didn't think so.
Hope this helps.
-Bob
