I'll just jump in here to say that audibility of low frequencies is a very complicated subject.
You can't really define any particular number for the lower limit such as 20 Hz because the threshold of audibility decreases with frequency. See, for example, this model of hearing sensitivity vs. frequency:
en.wikipedia.org/wiki/File:Lindos1.svg. This means that if you are doing a sine sweep or successive tones that decrease in frequency, you must increase SPL as you go lower or else they become harder to hear. Note that in the picture, the curves stop at 20 Hz, but in reality, the audibility threshold continues to increase for even lower frequencies. Note that the threshold level at 20 Hz (73ish dB) is still a long way away from a level that would be considered painful or even uncomfortable. We still have plenty of dynamic range in our hearing at that point. Our speakers? Probably not unless they are capable subwoofers.
Another thing that muddles the discussion a lot is tactile sensation. Tactile sensation is itself a very complicated subject. Vibrations induce a variety of feelings in the body, depending on the area of the body that's affected, the listening environment, and the content and frequencies involved. Vibrations can be transmitted into to your body through the furniture that you sit on, through your feat on the ground, and even directly between from sound itself. (Sound is just vibration of the air). The amount of energy that is transmitted depends on the physical properties of the materials involved and may even depend on where in the room you are standing, independently of SPL. The vibrations that enter the ground or your furniture may have also have travelled different paths to get there. It could be your subs are vibrating from their back-and-forth motion and transmitting vibrations to the floor directly, or the sound could be causing the floor or furinture to vibrate.
I mention all of this to illustrate just how complex subject tactile perception subject is. Audio purists may wish to dismiss tactile sensation as an "unwanted contribution" or at least an independent one. However, I don't think this is appropriate. If you only listen to headphone, then perhaps you can isolate the rest of your body from vibrations, but I believe headphones still do induce some tactile sensations. And anyway, you don't listen to a live event through headphones. In a live performance, the sound hits you, the floor, and your seats. It vibrates you, and you can feel it. (I'm thinking now of when I went to hear Kodo, a Japanese Taiko drum troupe perform live.) Totally acoustic, and incredibly tactile.
As for the frequencies involved. The truth is that both audibility and ability to perceive tactile vibrations drop as frequency drops. However, the body appears to be capable of perceiving vibration down to 1 Hz or even lower if the SPL is high enough. At the lower limits, the sensation is like being pushed by wind. Technically speaking, it's not really physically different from being pushed around by wind, either. Once you get to 10-20 Hz, some internal organs begin to resonate and the ears begin to perceive pressure. Your organs actually have mechano-receptors embedded in them to detect these vibrations. These can induce feelings of excitement or dread, a feeling of butterflies in the stomach. Evidently, our bodies are evolutionarily wired to be aroused by such sounds as they can warn as of potential danger. Above 15 Hz, our skin becomes more sensitive, and we can feel the vibrations more acutely. This trend continues to about 40 Hz or so at which point chest resonance starts to be a factor. Chest resonance is active from 40 Hz to much higher frequencies, at least 200 Hz may maybe even 500 Hz or more. I believe the skin's mechano-receptors are actually most sensitive at about 250 Hz! If you've heard good recordings on a a good live PA system in a large room or, especially, outdoors you are probably familiar with that wonderful super-tight chest bump you get from the kick. When you have a clean bass response and a lot of dynamics, you can tolerate a lot of transient SPL, and get music to kick the chest really good. Sadly, very few consumer HiFi-style speakers can provide this level of dynamics or sound way too harsh at the playback levels required, even with big amps. And then there is, of course, the loudness war that holds back the dynamics in music recordings, requiring you to have to play things "too loud" to get enough SPL for dynamic drums and bass. And furthermore, the acoustics in most small listening rooms totally destroy the bass and mid-range response of any system.
OK, so does tactile sensation dominate perception over hearing below 20 Hz? Lots of people think yes, but I think the jury is sort of out on that point. As I describe above, tactile sensation is: (1) a substantial part of the experience of a realistic acoustic event involving enough SPL, one that musically speaking spans genres from rock to electronic to jazz to classical; and (2) is very dependent on the listening environment as well as SPL. In other words, you can't really test each mechanism of perception in isolation. (Tests with headphones only really tell you about headphone listening.) I believe that the sensations of hearing and feeling actually merge to an extent in our perceptual processing. This only makes sense. We don't walk around with headphones on. The sound we hear is always interacting with our body, and our brain uses all the tools it has available to it to perceive this sound, so not just ears but mechano-receptors in our skin, bones, and internal organs.
In at least some experiments in enclosed spaces, listeners have been able to readily identify sounds down into the low single digits. However, there is anecdotal evidence that content < 10 Hz is very hard to perceive in rooms with a concrete slab floor, which is too stiff to substantially vibrate at low frequencies. However, content in the 10-20 Hz octave is very much perceivable. Content below 10 Hz is much more likely to be perceived if one is listening on a suspended wood floor, which will tend to vibrate much more readily. This suggests that 10 Hz may be a lower limit in terms of audibility and maybe also tactile perception when one is listening in isolation from sources of vibration other than the air itself. This limit may also be another matter of SPL and the fact that the dynamic headroom required to reproduce anything meaningful below 10 Hz is too high for this to be realized with any available content.
About that content. How much stuff below 20 Hz is there? In real life, there's no low frequency limit. Real life sounds are frequently full-bandwidth, and very often exhibit increasing energy as frequency decreases. So to an extent, nature has a tendency to correct for the roll-off in hearing sensitivity, but it really depends entirely on the sound. The most dramatic examples involve large scale phenomena such as wind, thunder, volcanic activity, and large flows of water. Also, very large rooms and caves can exhibit significant ultra low frequency resonance that can be excited with the right kind of stimulus such as wind or foot steps. To give a much more mundane but familiar example, you experience a lot of ULF when you drive with windows down, especially if one is a lot further down than the others. For home theater, ULF capability is pretty much a no-brainer if your budget and room placement constraints can accommodate it. ULF contributes a considerable amount of audio realism. Want to relax to the sound of the ocean waves rolling in? Better get some awesome subs, because those lowest frequencies do contribute to the listening experience.
As for music, ULF is probably a bit less critical, but it depends on genre and recording quality among other things. Live music PA systems typically extend to 40 Hz. Most commercial cinemas extend to 30 Hz. Only rarely do such systems extend any lower. This fact could be taken as evidence that < 30 Hz (or even < 40 Hz) doesn't matter for music. It is fair to argue that it matters less than say mid-range performance does. However, the primary reason these systems don't extend further is simply economic. Playing low frequency bass at high SPL in very large rooms or outdoors is expensive. It is also true that most musical instruments don't play below 30-40 Hz or so, but there are some that do. Larger organs are one example, but percussion instruments are probably the more common exception. Kick drum and especially large symphonic bass drums can have a substantial amount of ULF content. Then there's stuff like hall ambiance, which may exist independently from the sound being produced by the musical instruments. Some genres of music also have a lot more lower frequency content, including some kinds of electronic dance music and synthesized music. There's even a handful of stuff that's chock full of heavy ULF content on purpose. Personally, I think 20 Hz extension is worthwhile if you can accomplish it in your space, even if a majority of music falls off below 30 or 40 Hz. When the extra content is there, you'll know and appreciate it. That 20-30 Hz band can be a night and day difference with the right content, and having that extra bass capability can really distinguish your system.
I'll just wrap up with some more comments about hifi speakers and deep bass and a criticism of hifi system designs, in geral. Simply put, the bass extension of HiFi speakers will almost always be pathetic, and it will almost always be better to supplement them with subwoofers. Even what I'd call "high performance" (not "Hifi") speakers really have no business playing low bass. Part of the reason is that good low frequency performance requires a lot of power to achieve without having excessively large cabinets. This is popularly known as Hoffmann's Iron Law. The trade-off is between (1) extension; (2) efficiency; (3) box size. This trade-off is not absolute, as efficiency depends on other variables like motor strength and the type of alignment used, but these compromises still come into play. Something interesting and (in my opinion) tragic occurs in a majority of Hifi designs: In the effort to improve the measured extension of the speaker, some kind of signal shaping is used that destroys the upper-end efficiency of the system. This signal shaping is accomplished using a combination of passive and mechanical components. For example, mass may be added to a woofer to increase its output at the lower end of its pass-band but this compromises sensitivity. In a speaker design, the mid and tweeter are likely much more efficient than this woofer is, but in order for the speaker to have a flat frequency response, the sensitivity of these drivers must be padded down. It's not unusual for the majority of mid- and high frequency energy from the amp to be dissipated in a resistor and not even reaching the driver coils. The extra mass only hurts the efficiency of the woofer, so even though most of the amp power gets to the driver, more of it gets used to heat up the voice coil than in a driver with less mass. Apart from the fact that you'll need bigger amps to reach the same output (you'll need those for your subs anyway), you're also putting more heat into the voice coil. This can cause power compression, which robs dynamics, and flux modulation that causes nasty inter-modular distortion. When actually producing those lower frequencies at a higher level, those woofers will also be undergoing more excursion, which is a separate (and probably worse) cause of inter-modular distortion. So keep in mind when looking for speakers that there's no free lunch. If you want more extension, then you will give up sensitivity, low distortion, and realistic dynamics.
These problems all exist to an extent with subwoofers as well, but subwoofers are more immune to them. Subwoofer drivers often have much larger and longer voice coils. The increase in mass and inductance harms high frequency performance but often allows for much more excursion and provides a lot more cross-sectional area for heat to be dissipated through. The added mass can actually help for a lot of designs because they are often intended to be used entirely within the bottom octave or two of their pass-bands. Furthermore, the larger radiating area of higher diameter drivers helps to offset some of the efficiency loss. Furthermore, because the sub is handling a narrower range, any intermodular distortion will be less severe than in hifi woofers.
There is another key difference though. Most subs are active systems that often employ DSP to shape the response. The biggest drawback of passives, in my opinion, is that you can't electronically shape the response of the drivers independently. In the "pro" world, where very high output per dollar per pound is required, the drivers and systems are typically designed differently. Pro drivers are typically larger in diameter, lower in moving mass, and use stronger magnets than hifi drivers do. The greater diameter and lower moving mass improve sensitivity in the high frequencies. The stronger magnets are more expensive and heavier, but they improve the efficiency of the motor itself. As a side-effect of these changes, such drivers begin rolling off at considerably higher frequencies and tend to have relatively poor extension. This can be improved using a vented or horn alignment, but often some low-end droop persists. For a hifi buyer, that might be a major drawback to a speaker. Like, would you want a speaker that's 6 dB down by 200 Hz? That sounds like crap. However, in the pro world, systems are almost always active and there's electronic EQ involved. This means that the low-end response of whatever woofers are in-use can be boosted without killing efficiency, so they can be used to pound out mid-bass with little power compression. Furthermore, the mids and treble can use very efficient drivers without padding circuits, allowing them to play more dynamically. Of course, these systems do require extra power and more excursion to achieve the bass boost, but at least as far as the power is concerned, the stronger motor is a plus if the system is being used below it's natural pass-band, as is often the case here. Essentially, what happens is that even though the bass roll-off starts at a much higher frequency, the roll-off is a lot more gradual so that by the time you get to the lowest frequencies, you actually have more efficiency than for a driver with less motor strength and a lower roll-off.
Alright, yeah that was a bit heady, but it's my take. Hifi passives are just too big of a compromise for me to bother any more. My stereo speakers are two-wav, fully-active and bi-amped with 2 channels each from an XPA-5, using SEOS-15 horns and Acoustic Elegance TD-12M 12" woofers crossed at 950 Hz. They are around 98 dB/2.83V/1m sensitive and the woofers are 4 ohm nominal. For transients, they can probably hit the high 120s at 1 m if the signal is wide-bandwidth, but can probably do mid-bass at 120-123 dB @ 1 m. The ported alignment allows them to play down to about 40 Hz before dropping steeply, but their response is already drooping a bit below 150 Hz. Who cares! My room actually boosts those frequencies quite well, and in any case, my system has dedicated mid-bass woofers that begin to kick in well over 80 Hz.
And before you ask why I'm so crazy as to have a 120 dB capable system in my living room. It's not about loudness. It's about realism. Where I sit, SPL diminishes by 10 dB. So really, my system can hit ~110 dB per channel at the seats. Room effects mostly boost output but also harm output in places. With room treatment, some of this boost is diminished, so seating distance matters more. Then there's the need for DSP. How much headroom can you spare for DSP? If your preferred playbacklevel is a few dB below the point at which your speakers are objectionably distorting, then any DSP you throw on top for room correction will hurt your performance. You *need* headroom for room correction to work at its best. And in fact, there's evidence that as listeners we are more sensitive to frequency response anomalies caused by poor acoustics at *higher playback levels* where the speaker is least likely to faithfully reproduce the sound, as processed by the DSP. Lastly, 110 dB isn't that loud for realistic sound. Natural sounds can have transients with crest factors that are 20-30 dB or more above the average level. You might be listening to a quality, high dynamic range recording at 70 dB average. This may be enough to be immersed but not necessarily to prevent conversation, but this track might want to produce peaks at 90 or even 100 dB. It varies *a lot* by recording, and most recordings today have abysmal dynamic range with crest factors near 6 dB or so. But those recordings don't sound realistic no matter what. (They aren't necessarily bad from a musical standpoint, but this gets into the realm of aesthetics and so on.) At the same time, I do have plenty of very good musical recordings with crest factors that reach to 20 dB and beyond, and with a lively playback level of 80 dB or more, having the headroom to hit those crests really makes a big difference for realism. And finally, listening louder is a lot more enjoyable on a system that is playing well within its limits. You might think that you don't need high output because you don't like to listen loud, but if your system sounded much cleaner, more detailed, and more realistic, you might find higher listening levels to be both more comfortable and more enjoyable.
Alright, I've revealed I have a real bias against hifi-style equipment. If I'm disappointed in the new Emotiva speakers, it's only because they follow in the tradition of most hifi speakers: low efficiency. I can do far more with my system using an XPA-5 than I could with XPR-1s on those speakers, and mine would sound way better while doing it. No puny woofers are going to take 1000W of power without exhibit power compression almost instantly. Any such design can't compete with with a high-sensitivity pro-style design, period. The real drawbacks with such designs are that they are typically larger (wider baffles for bigger woofers), heavier (bigger magnets in the woofers), and frequently requires electronic signal shaping to get the most out of. (Incidentally, wider baffles can improve sound quality even though they are often aesthetically unacceptable.) I hate to say it, but I suspect that marketability comes into play here, big time. How many of you are comparing speakers by their rated extensions? Supposing the ratings were at least consistent, will 30 Hz extension *always* be better than 40 Hz extension? Oftentimes, the answer is no. If the speakers have large built-in woofers with huge amps and active DSP, then maybe that 30 Hz extension rating really means something, but if you're looking at some passive speakers with a couple 6" woofers that 30Hz extension is probably of little practical use (because 6.5" drivers can't move that much air) and only serve to hurt the performance of the rest of the design.
A few more notes. My Hsu VTF3-MK4 and VTF-15H subs do a great job with bass down to 16 Hz. I'm using them in combination with two Hsu MBM-12s mid bass modules. These units were about 1.5X as expensive as the rest of my speakers and amp combined (until I built my fronts). The drivers for my upgraded subwoofers are on order, and my budget looks proportionately quite similar. Good bass requires a hefty investment, but it's worth it in my opinion. I would suggest others consider a roughly similar ratio of cost for speakers+amps vs. subs+amps. One of my new sub amps does an honest 2 channels X 6000W, all channels driven. However, the design uses ultra-efficient drivers that will barely tap that power except when asked to reproduce frequencies below roughly 15 Hz. The Hsu Research VTF-15H/VTF-3 and similar designs by companies like SVS, PSA, and others that deserve mention fill kind of a sweet spot for those who want great bass performance with fairly modest concessions in terms of money and/or size. There are definitely subs that are more capable for more money, but these are probably sufficient for most music playback. For home theater, more sub capability is warranted if one has the budget and is pairing with good high-efficiency speakers that can keep up. The major struggle for good bass sound quality is dealing with room acoustics, which can really only be adequately dealt with using multiple subs and ideally some room correction DSP. Here again, more headroom is a good thing.
(Ok it was Butterworth...)
