I've invested a lot of time, as of late, to trying to determine the best way to calibrate an audio system. When calibrating for in-room response, it may be readily observed that a flat frequency response does not sound so good. The result is typically too bright indoors and lacks bass both indoors and out. Sadly, many authorities argue for calibrating systems this way and claim that this "reference" response is the correct response and is something the listener should "get used to" unless they want to alter the response to suit their preferences. Let me just say right here right now that this is B.S.
Numerous blind listening studies by Harman indicate that tonal balance preferences are actually quite universal. Harman's work indicates that subjective sound quality impression correlates strongly to the flatness of anechoic response and the smoothness and monotonicity (greater toward the bass) of the reflected and power responses. These preferences have also been shown to be consistent between trained and untrained listeners as well as between American and Japanese listeners, the latter of which shatters the stereotype that Japanese listeners prefer more treble and less bass than American listeners.
My current calibration approach is based on the assumption that the first arrival dominates the listener's perception of tonal balance. This is perhaps the simplistic explanation for why an anechoically flat speaker sounds best in a wide variety of listening rooms. My reasoning is that the human hearing system analyzes the in-room sound to separate the sound of the source from the acoustical effects, where possible. The latter point, "where possible", is key because in many situations (including bass, in general), the hearing system cannot distinguish the source independently from acoustical effects. For practical purposes, first arrival may be approximately determined from an impulse response measurement using the FDW feature in REW. A 1/3rd octave window works outstanding for my situation, but I haven't verified that this will work for everyone.
I calibrate so that my first arrival response is flat, except for adjustments at the top and bottom end. I recently experimented with in-air high frequency absorption as a model to use for defining a family of roll-off curves, allowing adjustments to be made as needed for different content. This experiment has been very successful, so I thought I'd try applying a similar model-based approach for determining bass lift, which is necessary for a balanced sound.
To understand the need for bass lift in the first arrival response, it's helpful to recall what the true reference is: an anechoically flat speaker. Virtually all monitors used in professional audio production are designed to be as flat on-axis as possible when measured anechoically, excepting the roll-off at the top that is often added, particularly to monitors intended for near-field use. What happens is that as frequency gets lower, reflections arrive too early to be distinguished from the first arrival sound. Instead, these earliest reflections combine with it, contributing bass lift to the first arrival sound. This is in addition to the bass lift observed in the full in-room response owing to both the perceived first arrival sound and the later reflections. As a consequence, even anechoically flat speakers benefit from some bass lift when used outdoors or in very large spaces, at least if the content was mastered in a smaller room as music usually is.
So I'll spare the details of my analysis as it mainly involves talking about a lot of numbers, which vary quite a bit even between recommended music monitoring environments. The typical recommendations are that the anechoically flat speakers be placed at least 5 feet from the rear or side walls, preferably in a room with a taller ceiling than the usual 8 feet that's common here in the states. Taking into account these recommendations and assuming a modest listening distance of up to 20 feet or so, I determined that the primary contribution to bass lift arises due to the floor reflection. The frequency region and magnitude of this lift will vary considerably depending on the height of the bass/mid driver(s) relative to the floor and the reflectivity of the surface. By my calculations, the lift due to floor bounce falls somewhere in the range between 100-500 Hz or so. A reasonable guess for magnitude is 3-4 dB, but it could be more in some circumstances. I haven't nailed down any sort of "best" settings to use for the widest variety of content, and indeed, the ability to adjust the center frequency and magnitude may be necessary to achieve optimal sound with a wide range of content. However, in my experimentation I find I get pretty good results with a +3 dB shelf centered at 130 Hz. Any higher than that, and I get noticeably too much mud with some content.
Some secondary contributors to first arrival bass lift arise due to the side walls, ceiling and rear wall. However, if the system setup follows recommended practice, these reflections won't contribute much first arrival boost until well into the sub range. Each of these reflections will also contribute less to the first arrival response than the floor bounce will. Right now, I'm running a +1.25 dB shelf at 52.5 Hz, and another +2 dB at 25 Hz. Note that most monitors, even the nicer ones, don't extend below 40 Hz or so. For content that contains substantial deep bass, subwoofers will usually be used during the monitoring process. This presents a significant complication because the subwoofer must be some how level matched to the main monitors, and the methods of doing this vary. However, as long as the sub is placed close to the monitors, an in-room frequency response level match ought to also provide a first arrival level match. If the sub itself is not EQed and is anechoic flat, then some additional first arrival bass rise is warranted. Below 20 Hz, I would say there is no clear guidance because almost any sub used for monitoring will begin to roll-off at that point, even as room gain (both first arrival and overall) begins to rise rapidly as well. I believe in the limit of low frequency, all room gain is essentially first arrival room gain, so it probably makes most sense to just keep things flat below 20 Hz, except for maybe a gradual roll-off toward the low extreme of the system to minimize phase delay.
For movies, the situation again varies depending on whether the mix is done at mid or far field in a dub-stage, at mid field in a smaller room (most consistent with music mastering practice), or in near field (also used for music sometimes). As the ratio of distance vs. height increases as would be expected in a dub-stage, the magnitude of the floor bounce will increase somewhat. Some additional floor bounce boost may be justified for dub stage mixes. However, side-wall and ceiling reflections may arrive too late to substantially contribute to first arrival sound, so it might be better to avoid much (if any) boost in the 40-50 Hz range. In the near-field situation the floor bounce may be diminished a bit more, and other room reflections may be a bit weaker, suggesting that less boost is appropriate overall.
Anyway, my work here is very preliminary. I haven't done nearly enough listening to convince myself that my settings for bass lift here are necessarily optimal. I am reasonably happy with the bass right now. Something to keep in mind is that the amount of bass lift may be somewhat less important than the shape of the lift. If the shape is wrong, it will sound unbalanced. Unfortunately, I'd bet that a lot of people including even many on this forum don't know what balanced bass sounds like. I admit, I'm not even certain of this, but I've learned to trust my preferences and am able to easily make adjustments that make things sound better or worse. For example, balancing 40-50 Hz with 90-120 Hz is crucial. Too little 40-50 Hz, and kick drum lacks weight and power. Too much, and it sounds muddy with little punch or impact. I'd be willing to bet that a lot of people run 40-50 Hz way too hot relative to 90-120 Hz and drown out all the punch. Yet, as noted above, 40-50 Hz should probably still be running 1-2 dB hotter than 90-120 Hz, when measured in the first arrival. In typical rooms, significantly more room gain may appear in the full in-room measurement than in the first arrival measurement.
Interestingly, the importance of first arrival may partly explain why near-field subs often sound better and feel more tactile. When placed near-field, the ratio of direct to reflected sound may increase compared to other placements, and as long as the user calibrates in-room response to the same level and target curve, the subs will have higher first arrival SPL where direct-to-reflected sound ratio is higher. As such, it may be possible to get good tactile punch from subs placed in other parts of the room, provided the system is calibrated in a manner more like I describe here. Of course, the near-field placements may still offer more headroom and less interference from acoustical effects, especially where high Q room modes may be present.
So anyway, I can't really give a definite prescription for calibrating bass response, but I can state with moderate confidence that the notion that in-room or even first arrival bass response should be flat for a correct listening experience is total rubbish.
Edit: too much 40-50 Hz relative to 90-120 Hz causes bass mud and lack of punch or impact. (These are pretty rough ranges, of course.)
Edited by SME, 02 February 2017 - 08:26 AM.