|The Science of Domestic Concert Hall Design|
by Ralph Glasgal
Ambiophonics, 2nd Edition
Replacing Stereophonics to Achieve Concert-Hall Realism
By Ralph Glasgal, Founder Ambiophonics Institute, Rockleigh, New Jersey www.ambiophonics.org
A Listening Room For Ambiophonics
Originally, I thought that, as in stereophonics, listening room reflections would be deleterious to Ambiophonics. But this is not the case. In Ambiophonics room treatment or correction is not normally required except for the lowbass region where, as in stereo, room modes are a problem. However, speaker response correction is highly desirable. The reasons for these conclusions are discussed below.
The three main pillars upon which Ambiophonics is constructed are the software or mechanical Ambiodipole, surround ambience convolution, and speaker correction. One can enjoy ambiophonics listening to two liitle PC speakers so obviously it is not necessary to have a special room. Fortunately, turning a family room, spare bedroom or rec room into an acoustically viable environment for a quality domestic concert hall or surround experience need not require a big budget, a building permit, or even a single carpenter. The trick is to understand what factors might degrade Ambiophonic realism and then do something about them. Again, although room treatment or correction is not essential to an Ambiophonic experience, the technology involved will be described below for those who opt to create the ideal listening environment.
In sterephonic reproduction where the speakers are far apart, room reflections have a delay comparable to the direct sound and so do effect localization and interfere with the functions of the pinna. In Ambiophonics however the speakers are quite close together and one usually sits closer to them. In this case the early reflections from the room are somewhat later than the direct sound and so the effects of room reflections are minimized and akin to the early reflections one would hear in a concert hall from other seats and heads. Furthermore, if one uses surround speakers to provide ambience derived from hall impulse responses, then the even longer delayed signals from these surround speakers swamp any short delay room reflections. This does not apply to the low bass where room modes can effect Ambiophonic reproduction as much as they do stereophonic. However, having the front speakers so close together makes it easier to do bass management using the room response correction software and processors now available from several sources.
While two speaker Ambiophonics is already a major improvement over the stereo triangle, the ultimate in reproduced realism depends on the collaboration between crosstalk free front and rear speaker pairs and additional surround speakers (if the performance is situated in a large space rather than the home listening room or small studio). In general, a truly exciting binaural hearing experience is the result of the ear-brain system not having to labor at resolving conflicting sets of acoustic cues: the concert hall (as presented by the playback system) on the one hand, and the local playback environment on the other. The less adulterated the set of cues, the more persuasive the experience. Again, in most Ambiophonic systems the effect of the local environment is much less audible than is the case in stereo or 5.1.
The possible causes of acoustic disappointment are many but, happily, experience shows that most home media rooms suffer from insufficient absorption. For realistic concert-hall like reproduction we must eliminate any characteristics of the home listening room that modify the rear hall and front stage we are going to create via convolution and the Ambiodipole. For Ambiophonic purposes it is only necessary to get the reverberation time of the room down to about 0.2 seconds, which is far from the .01 of an anechoic chamber. Remember that in a real concert hall there are some short early reflections from nearby seats or people and so a completely dead environment at home is not a requirement. However the trick is to deaden a room over the entire audible frequency range and this requires different techniques in the treble and in the bass. In brief we will see that treble reduction of early room reflections is best done using inexpensive wall treatment while bass reflections and room modes are best tamed with electronic speaker/room response correction systems. Again this is gilding the lily and except for the bass response, the improvement is likely to be inaudible to most listeners.
Sounds arrive at a listener's ears from many directions: from sources themselves (the speakers) and from walls and objects that reflect sound toward the listener, much as mirrors reflect light. Because reflected sounds must travel further, they arrive at the listener after the direct sound with an altered frequency response and loudness level. The brain interprets these reflections differently, depending on which direction they come from, on how much later they arrive, how they are tonally changed, and how much louder or softer they are. (Curiously, reflected sounds can sometimes be louder than the direct sound in small rooms if they take two or more paths to the listener-say from the ceiling, floor and a side wall-and if the path lengths are the same so that they are additive. A reflected sound that follows the direct sound by less than about one-fiftieth of a second is perceptually fused with the direct sound, i.e., the brain generally cannot distinguish the two as separate acoustic events. But despite this, uncontrolled, strong, and very early reflections (0 to 20 msec) make a mess of perceived tonal quality and wreak havoc with stereophonic (but seldom Ambiophonic) imaging. Reflections arriving somewhat later are interpreted as room ambience. Reflections trailing the direct sound by more than about one-fifteenth of a second can be heard as discrete echoes or more likely as reverberation. Shorter echoes can be particularly offensive if the room concentrates or focuses such sound. Concave room features, in general, such as bay windows, are frequent culprits and should be avoided if high-quality acoustic results are intended.
Room surfaces have three primary acoustical properties-absorption, reflection, and diffusion (a complex form of reflection)-but only absorption is of real use in the cause of eliminating audible room reflections at the listening position. Couches, carpets, cabinets, bookcases and other furnishings all contribute to a room's reflection patterns, albeit usually in unplanned and acoustically erratic ways. For example, carpeting on a concrete or hardwood floor soaks up a fair amount of treble energy, but allows bass to bounce right back into the room. Large closed glass windows typically reflect middle and high frequencies back into the room, but let bass pass right through. A bookcase might absorb highs, scatter (diffuse) mids, and ignore the bass altogether. Thus, a room for Ambiophonic listening should, if one is pursuing this option, be treated for a real reduction of wideband reflections.
More Evil That Rooms Do
While the ideal Ambiophonic loudspeaker would aim its sound only toward the listeners, most loudspeakers spread their output, to some extent, like floodlights illuminating both people and surroundings. A speaker firing directly at the listener will also direct sound sideways, up and down, even backwards. In a typical untreated room, this "unaimed" energy hits a wall or cabinet and bounces back toward the listener only a split second after the direct sound. Think of these delayed versions as the acoustical cousins of multi-path "ghosts" on a TV screen. But this is true of most all acoustical environments and the brain can cope with this if the field is binaural but not if it is stereophonic.
Thinking Ambiophonically, it is necessary to understand the following. The average untreated living room has a reverberation time of about six-tenths of a second. Since a recital hall could have a reverberation time of as little as eight-tenths of a second, and even a concert hall can be in the 1.5 second range, the typical home listening room reverberation time is surprisingly significant compared to the halls in which music is performed. Let us assume that we are playing a recording of a large choral work that includes a normal ratio of direct sound to hall reverberant pickup. When such a recording is played stereophonically in a typically small, live, home environment, the direct sound stimulates the listening room to produce a reverberant field that tells the brain that the performance is in a room that is small and bright. But then the recorded reverberant field reaches the ears and tells the brain that the room is large and acoustically warm. When you add to this the comb filtering and pinna effects due to the spurious directional early room reflections that further confuse the brain, it is no wonder that stereo playback of recordings of larger musical groupings never seem to be realistic no matter how much we tweak our systems.
In the Ambiophonic case, one normally has surround speakers generating ambience with reverberation times of around 2 seconds so if the room reverb time is reduced just by half to say .3 seconds the room effect becomes inaudible. Additionally with the front speakers close together one can be a bit closer and the ratio of direct sound to room reflections increases making the room effects on localization that much less significant.
Soaking It Up
Absorbers are devices designed to soak up sound. Most absorbers work by converting acoustical energy into thermal energy. Typically they do this by forcing sound waves through a dense maze of small fibers that rub together to produce friction and heat. Carpet, soft furnishings drapes and even clothing can provide useful absorption in the treble and upper midrange, where you'll find female vocals, violins, trumpets, flutes, cymbals, and other high pitched sounds.
Acousticians refer to special sound-soaking materials like fiberglass batts as frictional absorbers, or more colloquially, "fuzz". Generally, the thicker and denser the fuzz, the more effectively it traps sound. A dense, two-inch thick fiberglass panel mounted directly on a wall absorbs nearly 100% of sound incident upon it in the range from 500 Hz (about one octave above middle C on the piano) up to 20,000 Hz, the approximate upper limit of human hearing. To absorb much energy below 500 Hz requires a significantly thicker panel, usually 4 inches, or an air gap of a foot or two between the panel and the wall. Either way, using fuzz to soak up the lower midrange and bass requires considerable space. Devices such as resonating tubes, i.e. fuzz surrounding a tall tubular cavity are only marginally effective in the bass region. As we shall see the expense of such bass absorbing devices is much better invested in a computerized speaker/bass correction system.
Splayed Walls for Fanatics
If building a new listening room or remodeling and existing room. It is possible to splay both of the side walls and front and rear walls. The walls should lean outward at an angle of five degrees or more as they increase in height. The conventional wisdom has been that eliminating parallel surfaces is not worthwhile since the behavior of such a room in the bass frequency region is unpredictable in advance and hard to measure after the fact. But bass standing waves are not the only problem one must find a solution to and bass correction systems handle bass without difficulty even if the walls are splayed.
For upper midrange and high-frequency sounds the soundwaves coming from floor-standing loudspeakers will be reflected, as light would, in an upward direction. As these rays go from wall to wall they must go up to the ceiling before they can return to ear level. Hopefully, in making this longer up-and-down trip, they will lose significant energy and also fall beyond the critical 20-millisecond early reflection time zone. This is essentially a benign form of diffusion, which largely avoids diffusing sound to the listening position. In general, splaying the walls can make the absorption treatment of the walls and floor a little less critical.
The amount of absorption that should be placed in a room varies according to the room's size. All things being equal, a big room sounds more live than a small one, requiring more absorption to bring it down to the same level of acoustical merit. This quality is expressed as reverberation time: the amount of time it takes for a sound in a room to drop 60 decibels in level from the moment the source stops producing sound. The shorter the reverberation time, or T60 as it is called, the dryer the room sounds. In general, a dedicated Ambiophonic listening room should be quite dead with a reverb time of .3 seconds or less. Because it is derived by averaging the time it takes sound to decay by 60 decibels across a broad segment of the audible spectrum, describing a room with a single reverb time figure is often misleading. A poorly designed room might boast a textbook-perfect average of T60, yet sound disjointed and unpleasant because some frequencies die out quickly while others linger on and on. Ideally the T60 in any one-third-octave band between 250 and 4,000 Hz should not deviate from the average T60 by more than 25%. Translated into frequency-response terms familiar to audiophiles, this ensures that the room's reverberant sound energy is flat within about a decibel or so throughout the most sensitive range of human hearing.
One challenge lies in controlling reverberation in the bass frequencies where T60 figures might easily be triple or quadruple that in the midrange. Unfortunately, the use of fuzzy coated tubes, Helmholtz resonators and other well advertised gimmicks are largely ineffective. But if left unaddressed, the lack of low-frequency absorption can create an annoying unevenness in the reverberation characteristic of a home theater, media room or Ambiophonic home concert hall. We shall see below that electronic room correction systems are the answer to the Ambiophile's prayer.
The Weight of The Sabines
The Sabine is the unit of sound absorption and it is computed by multiplying the area of an absorbing surface in square feet by its absorption coefficient. The absorption coefficient is simply the fraction of sound that is absorbed by the material at a particular frequency or over a band of frequencies. Thus a window open to the outside swallows up any sound that passes through it and, therefore, has the highest possible absorption coefficient of one. If the open window is one- foot square, its total sound absorption is one Sabine. Ten square feet of 4-inch thick fiberglass could absorb some 9.5 Sabines at 500 Hz and higher, but only about 7 Sabines at 100 Hz. A 660-cubic-foot room (10x14x19) would need approximately 700 Sabines of absorption to get down to a reverberation time of .2 seconds. Using 4-inch fiber wall panels, the area requiring padding would be in excess of 700 square feet, or about half the surface of the room allowing for the small absorption contributed by other surfaces such as rugs, drapes and furniture.
Tacking up 100 square feet of fuzz on each sidewall yields the same absorptive value and produces the same T60 as moving the fuzz to the front and rear walls. However, the quality of the stereo sound you hear, even the intelligibility of music and dialog, could differ dramatically. Absorption is best deployed on the ceiling and the front portions of the side walls, where they prevent sound from the main front stereo speakers from bouncing into the listening area a split second after the arrival of the direct, speaker-to-listener sound. If left untreated, these reflective surfaces allow strong early reflections to disrupt tonal balance and imaging, and scramble the often subtle aural cues in stereo that give music and soundtracks their texture and life. However, the Ambiophonic situation is much less critical and one can just treat as many surfaces as one can bear to treat. As with most things in life, compromises may be necessary. Remember, even if your listening room is not Ambiophonically perfect, neither are most concert halls.
Sabine noted that halls exhibit the same basic sonic behavior at very low sound levels as at very high ones. If you are an active concertgoer, you may have noticed that concert halls show their distinctive sonic personalities even during those hushed moments when the maestro mounts the podium and raises his baton. Recreating in a residential setting the characteristic sound of a real hall begins with getting that "silence" right. Unfortunately, the typical home is neither designed nor constructed to allow the Ambiophile to hear the desirable level of sonic detail. If you turn off your playback system, shut the windows and door, and just listen to your listening room for a few minutes with eyes closed, you'll be aware of how much noise is there. Acousticians have developed a sort of numerical shorthand to describe background noise levels. Known as "noise criteria" (NC curves, and usually specified in increments of 5, from NC-70 (extremely noisy) down to NC-15 (very quiet). These curves are weighted to account for the fact that the ear is less sensitive to low frequencies than to high. The curves' numerical designations are arrived at by taking the arithmetic average of sound pressure levels at 1 kHz, 2 kHz, and 4 kHz. A useful target for a purpose built Ambiophonic listening room is NC-20; a spec often encountered in the design of professional recording studios. NC-35 would be the minimum standard for a legitimate Ambiophonic experience.
Unlike treating a room to reduce reflections, keeping outside noises outside is probably a job for an outside contractor as major structural alterations involving gypsum, studs and concrete are often required.
One of the most universally vexing problems of the home audio experience, stereo or Ambio, the fact that residentially sized rooms give erratic support to low-frequency sounds. When a particular bass note's wavelength precisely fits a major room dimension, the note is strongly reinforced or cancelled in a phenomenon called a standing wave. Bass will boom or fade depending on where one is in the room and the frequency involved. The room also exaggerates or cancels any higher harmonics of these low bass frequencies. However, as the absorption properties of the room begin to take their toll this standing wave effect fades. Basically, standing waves are due to the fact that most rooms simply can't attenuate bass reflections enough to prevent them from interfering with themselves over several rebounds. Or another way of stating the same thing is to observe that the T60 bass reverberation time of most small rooms is much larger than the treble T60 and that the density of this home tail is much greater than that found in the concert hall.
Eliminating bass modes is the subject of much quackery. There are magic room dimension ratios, which help a little, and there are the resonant boxes and tubes for room corners which help a little. But even a room that is painstakingly dimensioned and equipped with tubes galore to provide the smoothest possible distribution of low frequency modes will seem bass-boomy in some places, weak in others and about right somewhere else. Fortunately new Digital Signal Processing logic has come to the rescue to solve this problem with singular success and relatively low cost and simplicity.
Room Correction Systems
In essence we want the bass response of the room to be correct at the usual listening area of the room. It really doesn't matter much what is happening in the corners or behind us when we are not sitting there. So let us temporarily set up a microphone at the listening position or even several such adjacent positions and measure the bass characteristics of the room and the speakers (and the amplifiers for that matter) at that point. Once we know what the room and the speakers are doing to the bass we can get a digital-computing engine to correct any errors in bass response in both amplitude and time. A room correction system is essentially a very fine parametric equalizer able to control amplitude at any bass frequency (or treble for that matter) with a resolution of 2 Hz or better.The most exciting feature of an RCS is its ability to measure both the speaker response and the effects of the room on this response and do something about them. Once the peccadilloes of the speaker and room are known (by launching a series of test impulses through the system to the microphone and getting the impulse response of the setup), the room correction software can then calculate the fine grained amplitude and delay equalizer settings needed to eliminate them.
The methodology of measuring the impulse response rather than the frequency response has a tremendous advantage over conventional steady-state-tone measuring methods. Say one measured the bass loudspeaker/room response using a sinewave oscillator and a microphone attached to a meter. Then, using the resultant curve to set a conventional equalizer feeding the speaker, one would assume that a flat bass response would be achieved. Wrong! Music, in particular, consists mainly of transients. Thus, if a standing wave in the room causes say a loss of 10 dB at 100 Hz at the listening position and we apply a 10 dB boost at the speaker, then a brief but audible 10 dB peak will be heard until the standing wave room response catches up to cancel that peak. It is not the frequency response of the speaker/room system that needs to be corrected but the transient response. While improvements in this kind of room correction system will come at an increasing rate, present systems can only cancel early reflections within a period of one wavelength of the frequency involved. Thus, a reflection off the rear wall from ten feet behind the microphone will be delayed about 20 ms. This delay corresponds to the period of a frequency of about 50 Hz. Thus a typical room correction system will not be able to deal accurately with the components in this reflection at frequencies above this. On the other hand, bass corrections are quite effective for near reflections coming from the floor, ceiling or walls. It is providential that the electronic room correction systems work best where conventional absorption treatments work worst.
It is inevitable that room correction modules will be included not only in Ambiophonic processors but in stereo and video control centers as well. It is anticipated that before too long Ambiophonic processors will appear that include room correction, Ambiodipole software and the real hall convolution computer.