I'll lay my cards on the table before I launch into this post and say that, based on my reading of the science and my own experience, I believe this speaker has issues in the bass response and that I believe it's going to sound lean unless placed close to a wall (which is not something I would want to do, and which therefore would rule this speaker out for me).
Ok, what I'm now going to do is to look at the Harman research on loudspeaker preference (specifically as it relates to bass response), try to correlate that with measurements of this Harman speaker, and finally try to speculate as to why this speaker appears to diverge from the Harman recipe.
(1) What does the Harman research tell us about what a speaker's bass response should look like?
The obvious starting point here is Olive's well-known model (2004, 2005), laid out in his paper
here. The details barely need repeating on ASR, but suffice it to say that, based on statistical analysis of listener preference ratings and anechoic measurements of 70+ speakers, Olive concludes that 30.5% of listener preference is accounted for by bass extension, which is defined as how low in frequency the speaker's -6dB point is in relation to its mean listening window level between 300Hz and 10kHz.
His conclusion is unambiguous:
lower = more preferred.
Toole concurs with Olive that bass extension is the relevant metric but, on the basis of his
own work at the NRC in 1986, finds that it is in fact the -10dB point that correlates to listeners' perceived bass extension. In Sound Reproduction, he writes that:
Later
work by Olive et al investigated loudspeaker bass preference more specifically. They asked listeners to adjust the bass response of a Revel F208 in an acoustically optimised room using a tone control, until listeners arrived at their subjectively preferred relative bass level.
The mean preferred bass level is shown below (in black) relative to the predicted in-room response of the F208 based on Harman's spin data (in cyan):
View attachment 93275
We are in quite a good position here, because we can directly compare Harman's anechoic data and their derived PIR for the F208 with the preferred bass response from the study, and use it to infer what the preferred anechoic bass response of the F208 should be.
Below are those data. Using the difference between the cyan and black traces in the graph above, I've drawn (in red) an approximation showing what this speaker's response would be below 100Hz if it matched listeners' mean preferences in Olive's study (NB: I have used the Harman spin data here since it is from these data, not ASR's more accurate data, that the PIR used by Harman in the graph above is derived):
View attachment 93277
As is apparent, listeners appear in this study to have preferred bass that was not only extended, but indeed also slightly
boosted.
In other words, all published Harman and related research findings (of which I'm aware) suggest that
the ideal anechoic bass response in a floorstanding loudspeaker is flat and extended (if not actually boosted). There is no published finding suggesting that a rolled-off response, or a high -3dB, -6db or -10dB point, is preferred. And the one study that we have which correlates bass preference directly with a known set of anechoic measurements of a Harman floorstanding speaker (F208) tends to confirm this.
(2) How does this correlate with measurements of the F328Be?
Well, this is a relatively short section
The F328Be begins rolling off at around 80Hz, and has its -3dB, -6dB, and -10dB points at approximately 65Hz, 55Hz, and 38Hz, respectively. On top of all that, it seems to have a -2dB shelf below around 200Hz:
View attachment 93284
It's bass response diverges greatly from that specified by the Harman (and related) research.
(3) Why?
Just a quick treatment of all the possible options I can think of here:
(a)
The Harman research is wrong. Possible, but I know of no evidence in support of this, and of plenty of evidence to the contrary.
(b)
Kevin Voecks believes the Harman research is wrong. Quite possible, I think, and perhaps he has some (unpublished) research to back such a view up. It would also be consistent with what appears to be a trend of relatively "polite" bass with limited extension (compared to many competitors) across the entire Revel line (other than the Salon2). It's also worth noting that the -2dB shelf occurs approximately at the point at which the speaker transitions from 2pi to 4pi radiation.
(c)
The speaker is designed to be used with subwoofers. Also possible, although it would surprise me that the largest speaker in the Performa range were designed so that reasonable performance could not be obtained without subs.
(d)
The speaker is designed to be placed close to a wall. If this were the case, it would be odd that Revel doesn't explicitly state it (especially given they also manufacture on-wall speakers), and again given it's the largest speaker in the range.
(e)
Sensitivity was (hugely) prioritised over bass extension. Quite possible, although it would seem at odds with Harman's research findings, and the current situation (no pun intended) when it comes to modern amplification.
(f)
The enclosure is not large enough for the woofers. This actually seems increasingly like the most plausible explanation to me (albeit pretty bizarre given this is Revel we're talking about). We already know (based on independent measurements and published TS parameters) that the 8" woofers used in the rest of the Performa range need large enclosures to extend low in frequency. Could it be that Marketing told the design team they had to squeeze three woofers into an enclosure that in reality should fit only two?
Anyway, those are all the possible reasons I can think of....
