Back to REGONAUDIO

A revolution is about to happen. And I don't mean one of the small improvements or little steps forward, backward, or sideways that are often dubbed revolutionary in audio, at least by their progenitors. This is going to be a real revolution. In my opinion, this is going to be the most significant step forward in the reproduction of music in the home since the invention and practical realization of stereo: Using digital signal processing, it has become possible virtually to eliminate the adverse effects of the listening room on the sound of speakers, in effect, to cancel out the home listening environment and reproduce without interference the originally recorded sound. This has long been a dream. It is about to become a reality.

I have heard such a system in action.¹ Take my word for it: If first impressions count, this process is the ticket to a new level of sonic realism.

The process that corrects for room effects, which I am going to try to explain in a moment, falls under the rubric of digital signal processing (DSP). But so do many other forms of signal manipulation, and DSP speakers that do not actively address the room question are already appearing, and soon in large numbers, no doubt. To avoid confusion with the general idea of DSP, I shall refer to correction of room effects by digital processing as DRC (digital room correction) for the purposes of this article.

Perfection in audio comes from avoiding mistakes. And to understand how the DRC process accomplishes its' sonic wonders, we need first to review what speakers playing in a room do wrong---always and inevitably wrong, in fact.² Imagine a speaker that is perfect in its direct radiation "on axis"-no distortion, perfectly flat in frequency response, perfectly phase linear. What comes at you directly from the speaker is an exact replica of the electrical input to the speaker. Two of these speakers together are a perfect stereo speaker system-or would be if all you heard was direct radiation. But in your actual listening room, you are going to hear a lot more besides the direct radiation from the two speakers. And there lies the big problem.

Speakers always radiate in directions besides straight at the listener. Sound comes out of them off to the side, up, down, backwards, every which way. And all the sound bounces around your listening room and comes back to where you are-and back and back and back. For that matter, the sound that comes straight at you from the speaker goes by you and joins the procession of sound reflections. Audiophiles worry about the subtle ringing and time smear of cables and amplifiers. But that is indeed subtle compared to the time-smearing misbehavior of the listening room/speaker combination. This room interaction problem is the crux of speaker design.

The same observations apply to "live" sound sources in rooms. Everyone who plays a (movable) musical instrument will have noticed how remarkably different the sound is in different rooms, or compared to outdoors. This is easily noted, too, with the speaking voice if one listens for sound, not meaning. We all know the extreme "singing-in-the-shower" sound alteration phenomenon, but surprisingly large perceived sonic differences occur without such extreme differences in room acoustics.

The importance of the listening room and of the control of the speakers' interaction with the room is no big surprise nor secret. Everyone who has thought seriously about audio will have thought about this, and certainly all serious speaker designers have given the problem very careful thought, if with highly variable conclusions reached, ranging from the omnidirectional speaker as ideal to the ultra-directional.

Some of the ideas for dealing with the room interaction problem work out quite well in practice, though far from perfectly. In my opinion, the most successful approach to date has been via a combination of smooth power response (along with smooth flat on-axis response) and controlled directional behavior to minimize the early-arriving reflections from the side walls and the floor.³ In what I called "second revolution" speakers in my Snell B review a year ago, the control of radiation pattern makes possible remarkable levels of discrimination against room effects, especially from the midrange up. But even such speakers have serious limitations in overcoming the impact of the listening room in the lower midrange, mid bass and down, because it is not possible to produce highly directional radiation patterns in the lower frequencies and because it is difficult to make the room absorptive in the lower frequencies.

No doubt some of you are thinking, "Why is he talking so much about speakers, and so little about room treatments? I have my room so fixed up that all the problems are gone." Unfortunately, there are aspects of the room problem that cannot be fixed by acoustic treatments. Yes, indeed, you can soak up the higher frequencies. And absorbing/diffusing panels on the side walls will eliminate the early higher-frequency discrete reflections that confuse stereo imaging, among other adverse effects. But lower down, problems will remain. Think of it this way: Sound absorbing materials of the conventional sort have to have a thickness comparable to the wavelength to work at a given frequency. A 100 Hz note has a wavelength of about 11 feet. Clearly, we are not going to be able to fix up midbass with wall treatment very easily.⁴

The midbass and lower midrange frequencies are both audibly and measurably a real problem in actual listening rooms even in the relatively obvious sense of frequency response. Because of the (essentially undampable) reflections off the walls, floor, and ceiling, there are cancellations of the direct sound at some frequencies and reinforcements at others. Typical in-room, unaveraged frequency response measurements show wild ups and downs, peaks and valleys. And in the lower frequencies, these peaks and valleys occupy rather broad frequency bands. The graphs don't look anything like the nicely smooth anechoic responses of the speaker itself. And you aren't hearing anything like the speaker's nicely smooth anechoic response, either, unless you are listening outdoors up in a tree!

Magazines that publish measurements tend to publish measurements that make products look good, and a raw unaveraged over-frequency, unaveraged-over-space in- room response curve-which almost always I looks very bad---is a rarity. It is also a revelation. In the higher frequencies, it seems that the ear-brain does do some averaging over frequency bands and, via the small head movements that are always going on, over space; but in lower frequency ranges, the cancellation/reinforcement pattern does not usually vary rapidly with changes of positions, and small-head-movement averaging does not help. In those frequency ranges, the peaks and valleys can be most definitelyaudible.⁵

To put the situation with regard to spatial averaging in perspective, note that at 10 kHz, with wavelength around an inch, waves that are in phase at one position may be out of phase an inch or two away, and vice versa. Small head movements will do substantial averaging. At 100 Hz, wavelength 11feet, you would have to be Plastic Man to move your head far enough to matter. The crucial observation here is that you have to move half a wavelength to make a 180^o phase shift (in one wave).⁶ So the scale in space of reinforcement/cancellation patterns depends on the wavelength. In the midbass, we are talking about a big-scaled pattern, and spatial averaging on a small scale won't help. It really is true audibly, too, that real speakers in real rooms have lumpy, erratic midbass and lower midrange behavior. We get used to it because we have no choice. But that doesn't mean it doesn't matter. Unfortunately, it is easily and disturbingly audible.

The various room reflections do more damage than just making frequency response irregular. They also confuse imaging, smear fine detail, and, perhaps worst of all, impose the acoustic signature of the home listening upon the originally recorded sound. Very little of the music we listen to has its natural home in a room the size of a living room. Most music lives naturally in a much larger space; be it symphonic music, solo piano on a concert grand, rock music, even a solo voice of operatic style-almost everything wants a larger sense of space. We really want to hear the original acoustics, not our own living room.

Now that we have reviewed the problem, let's look at the solution, or what seems likely to be almost a complete solution. Since perfect stereo is the sum of two perfect monos,⁷ we can just think about one speaker. And, at least for one listening position there really is something that can be done to fix things. It is even something fairly obvious on a theoretical level, though the practical execution is trickier. What we need to do is to add correction signals to the direct radiation.

Suppose, for example, that the listening position is getting a big sidewall reflection after seven milliseconds. We just put in a correction signal, to arrive directly from the speaker after seven milliseconds and to be exactly 180^o out of phase with the wall reflection that we want to eliminate. So the wall reflection and the correction arrive simultaneously but exactly opposite in phase, i.e., one is the negative of the other, so they cancel out. It's as if the wall reflection never happened!

But what about the wall reflection of the correction signal itself, which will be coming along seven milliseconds after the direct arrival of the correction signal, or 14 milliseconds after the original direct signal? Well, yes, we have to have another correction for that, and of course, a correction for the wall reflection of that correction, etc., etc. "Big fleas have little fleas set on their backs to bite 'em... and so on ad infinitum."

This sounds nightmarishly complicated in description, but it is everyday bread and butter to mathematicians. As long as the wall reflection is lower level than the original signal, the repeated correction process converges to something nicely computable and definite. (For the techno-people, this is in effect just the convergence of geometric series with a complex ratio less than one in absolute value.) Of course, we want to correct/cancel all the reflections from walls, floor, and ceiling, or at least all the ones that occur early or with fairly high level. But the principle for doing this remains the same.

The figures here illustrate the process in the case of an impulse followed by a single reflection. ( These are from the AES preprint (3375 E-3 ) of SigTech designer Ron Generaux "Adaptive Filters for Loudspeakers and Rooms " .) The minimum phase EQ filter with frequency response shown in the third figure converts the impulse-plus-reflection shown in the first figure into the pure impulse shown in the second. The right EQ gets rid of reflections !

By the very nature of this whole business, room correction has to be different for different rooms-only the first floor reflection is more or less uniform from room to room in timing, if not otherwise, and even there the timing will vary with listening position, and everything else is highly variable in all respects. So somehow the listening room's own acoustics must be measured and taken into account. The natural way to do this is to put the speaker(s) in place, put a microphone in the listening position, and measure with the microphone what the room does to test signals from the speakers. At first sight, it might seem necessary to use a great many different test signals. But here the fact that the whole setup is what is called a linear system comes to the rescue.

Recall that a system is called linear if the output from the sum of two inputs equals the sum of the outputs from each of the inputs separately. What a room will do to speaker sound is a highly linear process, for any plausible volume, anyway. So if the speaker itself is highly linear⁸, the whole room-speaker combination is a linear system. And --here's the punchline--to measure a linear system, you can use just one single test signal. A linear system's behavior is completely determined by what it does to a pulse signal. So all one has to do is figure out the correction to make the impulse response of the speaker in the room perfect at the listening position, and the same linear correction process will make everything else, every kind of input, perfectly reproduced at the listening position, too.

So far, this is just mathematical theory. And in practice, things are a little more complicated. The pattern of the computation of the room correction, the "program" in the computer sense, is determinable by the one-time-only impulse response measurement of room behavior. This determination of the "program" can be done by a dealer-owned piece of (portable) equipment, and it wouldn't have to be done again unless you changed speakers or moved things-speakers, your listening position, large pieces of furniture-around your room. Then the "program" has to keep computing what to do for correction of the music signal as the music happens. Fortunately, audio is a very slow moving signal by the standards of computing nowadays. So it is possible to do the computations called for by the program in "real time," as the music plays. Also, fortunately, the processor needed to execute the program is a much simpler and less expensive item than the impulse-response-measurer/program determiner. So the whole thing won't be terribly expensive.

It is important to distinguish this whole DRC room correction process from old fashioned analogue equalization. For many years, purveyors of equalizers have claimed that you could adapt your speakers to your listening room using their devices. This did not really work, and a certain disillusionment has perhaps set in. But if you think about the situation as outlined already, the reason the analogue EQ devices failed where the digital ones can succeed becomes clear: The whole business of canceling room reflections involves time (or, equivalently, phase). If the would-be correction for some reflection arrives at the wrong time, it may well reinforce the reflection, not cancel it. The correction has to be opposite in sign from what is being corrected; the correction must have a compression where the thing corrected has a rarefaction and vice versa. In other words, at each frequency, the correction must be 180^o out of phase with what is to be corrected/canceled. Otherwise, if they are in phase, say, they don't cancel, they add together and things get worse, not better.

Now the analogue equalizers of the old days-or, for that matter, those of today-tie amplitude control and phase/time behavior together. What you do to frequency response determines what happens to phase. You can't control amplitude (frequency) response and phase response separately. But more importantly and in practice, analogue EQ devices cannot offer enough "coefficients", enough bands of EQ.

It is yet another one of those weird mathematical facts that any linear system can be described by its amplitude response (as a function of frequency) and its phase response (also as a function of frequency). For room correction, you need to control both kinds of response independently, in principle. And most especially you need to be able to have fine resolution in the frequency domain, with many bands of EQ per octave. This kind of thing can be accomplished quite easily by digital calculations. But it is essentially impossible with the usual analogue EO techniques.

I have been talking about the room correction process from the viewpoint of the speaker itself being perfect in its direct radiation. But that isn't actually necessary. The same kind of correction can be applied to the speaker itself, just as long as it is linear, i.e., as long as it has low distortion. In fact, that correction of the speaker's direct radiation is actually much less complicated. And digitally EQ'ed and phase-corrected speakers, which are already appearing, will no doubt be the norm soon. Such speakers will potentially offer remarkably good performance as far as on-axis direct radiation is concerned, but then so do some speakers now. It is really in the room correction process that the prospect for revolutionary improvement lies.

The comments that follow are based on only a two-hour listening session, provided by Kevin Voecks of Snell Acoustics, using a professional DRC system, the AEC1000 from SigTech applied to Snell Type B speakers. Although time was limited-I was about to leave for Denmark-the conditions were nearly ideal: I am intimately familiar with the speakers from my reviewing [Issues 73 and 75/76]. I am familiar with the room used, Voecks's living room, but not so familiar as to be habituated to it. The demo setup allowed me to use material of my own choice (CD only, though), and I had brought a number of CDs that I know well. And the setup allowed me to switch the DRC in or out at will, via a control at the listening position. So, though the following description is a first impression, it is one derived under the best possible conditions.

Since I am about to tell you how the DRC system improved the sound of the Type Bs, I need to remind you first that in my view, the Snell Type B is already one of the best speakers ever. In particular, I found in my reviewing that the Type B is one of the least colored, the very least colored, speakers I have encountered. The tonal truth of these speakers makes most others sound like coloration machines. Moreover, the Type Bs' design makes them much less room sensitive than most. As speakers go, the Type Bs need as little correction, room-related or otherwise, as any others, and far less than most. This point is important: Improving a bad speaker might be easy, but improving the Type Bs is improving some of the best.

Let me not try for suspense: The DRC process worked. The sound was improved in ways that were both audibly obvious and musically significant. There were big improvements, in fact. These were not things that would be hard to detect statistically in double-blind tests, that would take a week of listening to lock onto, or that would be noticed only by an experienced listener. I think any musically sensitive person would notice and appreciate the improvements at once and would be able to identify the corrected versus the uncorrected signal without difficulty. And, at least under the conditions I experienced, the corrected sound was clearly superior in musical terms.

The most obvious change that the DRC process made was in the mid bass. For reasons already indicated, midbass performance in home listening rooms is unpredictable except for being almost always quite bad. In the setup I was auditioning, there was an audible (and measurable) hole in midbass response, even though the Type Bs are not themselves at all attenuated in this region and in spite of the fact that the speakers had been placed in an optimal position in a room that is itself quite acceptable as listening rooms go. Problems of this sort just go with the territory of speakers in rooms.( Look back at the uncorrected Snell B in my room.)

The DRC process filled in the hole and smoothed out the midbass region. At the same time, it increased the midbass definition. Switching back to the uncorrected sound revealed how severely the midbass was suffering from room effects. No change of amplifier, cable, or CD player could have begun to alter the midbass enough to solve the room-induced problems; the DRC did.

Midbass balance and precision is an important aspect of sound as it affects music. In orchestral music in particular, this frequency range typically carries the whole harmonic foundation of the full orchestral sound. To get midbass wrong is to have the sonic fabric misaligned and unbalanced. To get it right, as the room correction process did, restores the balance and coherence of orchestral sound.

It is important again to realize that equalization in the usual analogue form, e.g., one-third octave bands, could not accomplish the result I have been describing. In my experience, the usual EO devices simply are not detailed enough to smooth out the perceived response.

It is important again to realize that equalization in the usual analogue form, e.g., one-third octave bands, could not accomplish the result I have been describing. In my experience, the usual EO devices simply are not detailed enough to smooth out the perceived response.

One learns to hear around one's own listening environment and as HP and the rest of us have often noted, good recordings can often capture the sonic signatures of halls remarkably well, with playback systems of high resolution enhancing the possibility of hearing the hall. But learning to hear around one's own room is just the point I want to make explicit here. A learned pattern of inference about the original acoustics is not the same thing as literally hearing the original sound. It has seemed to me that nearfield and/or an extreme kind of room treatment (LEDE, for example) were previously the only real possibilities for hearing the literal original acoustics. But the DRC process now seems to offer this kind of access to the original sound as a much more realistic possibility in a domestic environment. And, as already noted, in the lower frequencies say, lower midrange downward-the DRC process offers more access to the original acoustics than even near-field in a treated room would.

At this point in a typical review, all the good news would be followed by some bad news, by some sonic prices to be paid for the benefits gained. But here, there isn't any bad news, at least as far as I could tell. Within the context of CD standard digital, every change I heard from the DRC process seemed to be an improvement. Longer-term listening potentially might reveal problems, and the system has some intrinsic limitations that I shall discuss momentarily. But if it stops short of perfection, all its changes to the sound seemed at first listen to be for the better.

Listeners to analogue will have to decide whether the DRC improvements are enough to warrant digitization of the analogue signal--the DRC process requires a digital signal to work with. This decision I had no way even to begin making, not having an analogue source to try at the time. Note, however, that the ORC process itself, as opposed to any particular processor, is not tied to any specific digital standard. There is no reason why it should not eventually be available clocked to a higher sampling rate and/or more bits per sample than the CD standard. As things happen in digital technology-Le., fast-we shouldn't have long to wait for improvements if improvements are needed.

High End audio has a long history of skepticism about signal processing of any kind. Short, clean signal paths are almost an article of faith with the audiophile purist. Sometimes this attitude can reach surprising extremes; the idea of playing back the all too-numerous processed-to-death commercial recordings via a short and uncluttered signal path seems almost oxymoronic. But the attitude remains, that when it comes to handling the signal on playback, less really is more.

The digital realm requires rethinking of this attitude, however appropriate it may have been for analogue. It is certainly possible to do a bad job of digital signal processing; there were many reports of problems with early digital editors, for example. But in principle, if the digital signal is regenerated at the end in a clean, well-timed way, then digital manipulation of the signal will not have caused any damage, beyond whatever damage occurred from A to D in the first place.

Indeed, to my mind, it is precisely the possibility of manipulating the signal without damage that makes digital interesting. CD was not a sonic improvement in principle over analogue, however much more convenient it might have been. If you don't want to do anything much to the signal, you would be as well off to leave it analogue-maybe better off. But digital processing offers the possibility of doing signal manipulations that would be difficult or almost impossible in the analogue domain. In particular, digital signal processing makes possible the whole room and speaker correction process I have been discussing. And it makes it possible without introducing any degradation of the signal, at least in principle.

Speakers with digital room correction seem to me to be clearly the wave of the future, but one has to recognize that the DRC systems have certain intrinsic limitations. The possibility of digital correction will change the design criteria for speakers to some extent, but the DRC processors can correct only certain things, and some speakers will still be better than others.

First of all, the whole process depends upon the speakers being linear themselves, i.e., upon low distortion in the speaker. Low distortion has always been a significant goal in speaker design, but the wide variability of frequency response in speakers has tended to make frequency response the most obvious feature of speakers. Since frequency response is easily manipulated digitally, literal, on-axis flat response will become simply a non-issue. Nonlinearities, which are not susceptible to linear digital correction, will come to the fore as a major problem. And low distortion will become a major criterion.

A second limitation of the DRC process is that it does not really apply to the high frequencies on the practical level. Because high frequencies have very short wavelengths, the phase relationship between direct and reflected sound is likely to be very sensitive to listening position. Unless one wants to be a "clamped head" listener, it probably isn't practical to try to cancel the high frequency part of room reflections, above, say, 1-2 kHz. Thus one is left with the problem of attenuating the high frequency parts of the reflections by other means.

Fortunately, narrow radiation patterns are easy to attain in the highs, and wall treatment is also easier and more effective in the highs. Things fit together nicely here: The DRC process works well at the frequencies where effective room treatment is difficult or impossible, while ordinary, acoustic, room treatment techniques work well at the higher frequencies where the efficacy of the ORC process in literally canceling reflections is limited.

The DRC process can be used to provide a flat response in the highs in the sense of a spatial average in the vicinity of the listener. This may turn out to be enough, since there is good reason to believe that the ear-brain itself does this kind of spatial averaging via small head movement. (Otherwise, the perceived brightness of sound sources in rooms would tend to undergo extreme changes as one moved small distances.) But the detailed psychoacoustics of this part of the picture remain to be clarified by experiments.

A third consideration is that the DRC process, like most correction processes, will probably work the best when it has the least to correct. Also, one might expect it to work best when what it is correcting is a relatively smooth function of frequency. Taking for granted that the speaker itself is (digitally EQ'ed to be) on-axis flat and phase linear-and supposing that the speaker itself is, essentially, a linear system-then two considerations seem natural: First, that the speaker should have directivity that varies smoothly with frequency, and second that the speaker's radiation pattern should narrow at frequencies high enough that listening position becomes critical for reflection cancellation. (Incidentally, the Snell Bs that I was listening to satisfy these criteria unusually well, so I was admittedly hearing the DRC process under favorable conditions.)

Radiation pattern has been playing an increasing role in my thinking about speakers in recent years; I have come to the conclusion that it is a matter of surprisingly great importance in general, and perhaps I have been a bit obsessive about saying so although I have certainly not been either first nor alone in believing so. But in DRC speakers, it will be an obviously significant feature, indeed really the only feature other than low distortion.

One should note, too, that there is the possibility of user-controlled equalization in all this. But to my mind this is a side issue compared to the possibility of true sonic accuracy, independent of room effects.

I would like to turn aside from all the acoustic and linear systems theory, and say some things about the DRC process just in terms of my personal view of music. Since one cannot even begin to decide musical aesthetics for other people, you should just take these last remarks as one person's musical reactions.

I tend to take a literalist view of reproduced music. That is, I tend to compare what I am hearing directly to my recollections of live music with minimal accession of verbal categories. Of course, describing the results of such comparisons has to be done in words. But I believe the comparison itself should be done directly; and in my experience audiophiles sometimes use their audiophile vocabulary to justify sound that is indeed far removed from reality. (This tends to happen especially in regard to timbre and tonal balance, which are hard to express in words and essentially impossible to remember verbally.) Now High End audio has made remarkable progress in some directions. In particular, the nonresonant behavior and intrinsic clarity of present-day speakers is very impressive compared to most of the speakers from yesteryear. But certain definite failings in direct comparison with music remain. In summary terms, it seems to me that reproduced music isn't clear enough relative to its perspective, it isn't big enough in spatial terms (again relative to its perspective), and it is not correctly balanced tonally, or at least not reliably so.

Let me expand these points a bit. First, clarity: Close-miked reproduced music often sounds very clear indeed, but music recorded from an audience location tends to turn mushy and over-reverberant in playback and to be much less clear than it would be in reality. This unfortunate fact has led to almost all recordings having a closer perspective than is natural from the viewpoint of live listening. (Vocal recordings are especially frequent offenders here: in a real opera, Pavarotti doesn't sing to you from three feet away.)

Second, for all the progress in soundstaging, large-scaled music remains too small, spatially, in its reproduction. More precisely, it remains too small, given the perspective used. From the close-up, conductor's perspective (or very near it) that is used for most orchestral recordings, the orchestral image should be 50 or 60 feet wide and, say, 20 to 30 feet deep, at least. This image size essentially never happens in literal terms. In recent years, the abusive multimiking that used to make spatial nonsense of recording has become comparatively uncommon. But the size vs. perspective issue remains. Associated to this is a general largeness of size in the hall sense. As noted earlier, halls can be acoustically recognizable on good recordings, but their literal, huge size is seldom reproduced.

Third, not even "neutral" speakers are completely neutral in actual listening rooms. Meeting, say, +-5 dB limits would be impressive for in-room response. But no conductor would settle for +-5 dB balance among sections, and how would a string quartet sound with the first violin 10 dB too loud compared to the cello, or vice versa? This is not to mention the shifts of timbre in individual instruments attached to such imbalances.

All three of these major problems are associated to the negative influence of the listening room. Reflections impede clarity. The sonic spatial signature of the home listening room interferes with the sense of the original space and blurs stereo. And the listening room causes big, unpredictable irregularities in response, especially in the lower frequencies.

Look at these before-and-after graphs (not of the Snells) here: this is the uncorrected and corrected behavior of a different well-known audiophile speaker. The channel match ( as well as frequency response ) is much improved by the correction process and this is quite typical. Very few real-world systems exhibit truly symmetric behavior in-room, with a perfect channel match. SigTech to the rescue !

For all the progress in High End audio, progress in dealing with the room problem at the fundamental level has been limited. Some speaker designers have done what can be done via controlled radiation patterns, and to good effect. But most speaker designers, even High Enders, haven't really addressed the room problem comprehensively; though of course everyone is well aware that the problem exists, designers have seemingly felt that the additional complexity required was not justified. If one wanted to be abrupt, one could say that most High End speakers are just mid-fi speakers with better parts. I don't want to be so extreme. Years of speaker reviewing have left me somewhat frustrated here, in the extreme audibility of room interferences in most designs, and their residual audibility in even the best.

It is early to make definite judgments, but my first impression is that the DRC process offers the prospect of a truly fundamental step forward, fundamental not only theoretically but-what is incomparably more important-musically. Perhaps it will take some time for the practical execution of the electronics and/or the digital standard to rise to the level of ultimate refinement. But at the very least, the DRC process points up emphatically how far we are now from sonic perfection at the same time that it points the way toward the goal of the sound of real music in real space.

¹ Courtesy of Kevin Voecks of Snell Acoustics.

² Since this is not a product review as such, but rather an introduction to how a new kind of product will work, I am taking the liberty of describing how it should work in theory first, before I get to what it actually sounds like (of which I have only a first impression, in any case).

³ I have written about this so often that longterm readers probably don't want to go through it again. But if you want a run-through, cf., my Quad ESL-63 U.S. Monitor [Issue 73J, Snell B [Issue 75/76J, and Gradient 1.3 [Issue 77J reviews and my general article "How to Listen to Stereo" [Issue 64].

⁴ Actually, there is a remarkable product developed in Denmark at the Danish Technical University called SLAM that is only a few inches thick but absorbs bass frequencies. It works by controlled flexing of its surface rather than by absorption in the usual sense. While it is remarkably useful in controlling low bass reverberation, it doesn1 really address the problem I am describing, nor is it designed to, because addressing that would require absorbing frequencies right up through the midrange, while SLAM is designed to work at low frequencies.

⁵ I should point out here the important work, which has been surprisingly unsung in the audio press, of Roy Allison and Allison Acoustics on this subject.

⁶ Pursuing this further, if two waves are meeting at, say, a 90' angle and are 180' out of phase at some point, then the nearest point at which they are in phase is 0.35 wavelengths away. If they meet at a more oblique angle, closer to 0', then the in-phase point is further away (infinitely far if the waves have the same direction). If the waves meet at an angle near 180', the in-phase point is closer, with minimum distance, when the waves meet at 180', or one-quarter (0.25) wavelength.

⁷ This is in theory. In listening critical practice, stereo reveals things that would be hard to infer from the mono experience.

⁸ This doesn't mean flat in frequency response. Linearity has to do only with the absence of distortion, and a system can be completely linear and still have irregular, indeed arbitrary, frequency and phase response. Using the word "linear" to mean flat is really a malapropism.

Back to REGONAUDIO