Room Reflections & Human Adaptation for Small Room Acoustics

originally published: 4/4/16

This article is being republished with the addition of a related YouTube discussion on the topic of small room acoustics.

Some reflected sound is good. Sometimes a lot of the right kind is even better. Concert halls are deliberately reflective, highly reverberant, spaces. This is my “classical” listening room in our custom-built Canadian home. Conceived as a space for enjoying large, spatially involving, works of music, it was the largest “concert hall” I could afford at the time. The very neutral, essentially omnidirectional, Mirage M1s “became” the orchestra and the room became a seamless extension of the recorded space. It provided a very satisfying, involving, experience. Because of the designed-in irregular scattering surfaces, the heavy carpet and thick felt underlay brought the reverberation time down to under 0.5s so the room sounded much less “live” than one would think. It was a nice-sounding space, pleasant to be in. Late at night I have been known to sit in the dark with a glass of good Scotch and listen to non-classical involving pieces of music like Dire Straits “Brothers in Arms” played at high level. I miss this room. Elsewhere in the house was a 7-channel home theater with very few reflections but a very good multichannel upmixer and spatial synthesizer, a Lexicon CP-1 – this was 1988.

The article recently published on this site “History of Multi-Sub & Sound Field Management (SFM) for Small Room Acoustics” addressed room resonance problems below what we call the transition, or in large rooms the Schroeder, frequency. In domestic rooms it is around 200-300 Hz.  Below that frequency, the room dominates the quality of sound because of resonances. Above that frequency it is the combination of the loudspeaker axial frequency response and directivity, and reflectivity at the points of early reflections (ie. sidewalls and ceiling/floor) that are principal determinants. The following is based on content in my book, “Sound Reproduction: the acoustics and psychoacoustics of loudspeakers and rooms”, Focal Press, 2008, and all numbered figure references are to it. Not included in either discussion is the matter of adjacent boundary effects. These cannot be ignored in the mounting and placement of loudspeakers in, on, or near to room boundaries. These are discussed in Chapter 12 of my book.

Small Room Acoustics YouTube Discussion 10/15/19

Reflections below the transition frequency

The preceding article addressed control of room resonances that exist at specific frequencies where reflections between and among room boundaries combine in a strongly constructive manner. The frequencies are determined by room size and geometry, and in rectangular rooms they are easily predicted. In normal domestic listening rooms it is unusual to have excessive general "reverberation" at low frequencies, which is what is involved with reflections at frequencies that do not contribute to room modes and the associated standing waves. This is because the room boundaries and furnishings generally provide sufficient scattering and absorption. Usually, the "booms" are the problem, not the uncorrelated reflections. Often reverberation times measured at low frequencies are the decays of a few under-damped room modes. This is not reverberation; this is ringing!

The notable exceptions to this generalization are rooms with very reflective, concrete or masonry floor and/or walls as I described earlier in the serious listening room I set up at the National Research Council of Canada. Lots of basement rooms, and masonry rooms in steamy climates have excessive reflectivity at low frequencies, and not surprisingly, very energetic room resonances. In high-humidity regions it is good to have alternatives to fibrous or fabric absorbers.

In 2015 I published a 30-page paper in the Journal of the Audio Engineering Society, "The Measurement and Calibration of Sound Reproducing Systems" in the July/Aug issue. It is "open access" so it is free for all to download. Go to www.aes.org, click on publications, click on open access, type “Toole” and download. In it you will find an extensive discussion of bass buildup and the consequences it has to what we hear in homes and cinemas. Most people talk of high-frequency rolloff in rooms when it really is a bass buildup. They sound similar. A certain amount of it is unavoidable, and absolutely natural – it happens with all “live” sounds, including conversational speech. It is part of the sound of a room, something the ears expect if the eyes see a room. Too much is not good, and too little is equally undesirable: listening in an anechoic chamber is not pleasant. The key is to have the right amount of broadband absorption, and find ways to tame frequency-specific resonances without making the room overly dead.

Reflections Above the Transition Frequency

I rarely participate in internet forums of any kind, but I do look in from time to time. Occasionally my name appears, along with expressions of what people think I believe about certain things. I make an effort to ensure that anything I write or say reflects the results of accurate measurements and double-blind tests done by me or someone else. These are not personal beliefs, but the responses of numerous listeners, which may or may not have included me; most did not. Some of the investigations I refer to in my book were done in as geographically disparate places as Japan and Germany, so even “culture” is embraced. I wrote the words, but the data being reported are as neutral and impersonal as possible.

Reflections within listening rooms are real and numerous. Some would argue that they all are problems to be eliminated. Others take a more philosophical view that they just provide information about the room, and the brain can figure it out. I’m somewhere in the middle, but leaning towards the latter. The science that has been done so far seems to be on my side.

Figure 7.1: 2 vs 3 - Channel Fronts

Figure 7.1 above shows the first reflections in a stereo setup, and in an LCR arrangement. Anyone claiming that a phantom center image is superior to a real center loudspeaker has some persuading to do. The phantom-image situation is significantly muddled, and most listening situations are not perfectly symmetrical. As we will see later, eliminating all of the reflections does not solve the fundamental problem with the phantom center; in fact it makes it worse.

The contentious issue of side-wall reflections is the present topic.  An assertion of some participants in the audio forums is that I am in favor of these lateral reflections in listening rooms. According to some critics, I believe this in apparent ignorance of or dismissal of decades of professional audio “tradition” in which at least the first sidewall reflections are absorbed. Believe me, I am very aware of these points of view, having designed a few recording studios early in my research-scientist career, and also the high-power monitor loudspeakers that went into a couple of them (“Hi Fidelity in the Control Room – Why Not”, db – The Sound Engineering Magazine, Feb. 1979). In those days the main monitor loudspeakers were moderately directional (mid- and high-frequency horns) and sidewalls were usually angled to direct the first lateral reflections into the absorptive/diffusive back wall. It was clear that mixers preferred to be in a strong direct sound field and that is what they got.

However, as a research scientist, I saw a number of interesting questions to be answered, and so, it turned out, had other investigators. We needed specifics about the audible effects of reflections, and how to quantify them so that we can measure a loudspeaker/room combination and anticipate a subjective impression. We also were interested in whether the requirements for mixing a recording were the same as the requirements for enjoying that recording at home. All of this is discussed in detail over several chapters of my book.

Chapter 6 shows that in normal rooms the first lateral reflections in rectangular rooms of normal listening and control room dimensions are above the threshold of audibility.  They can be heard, but are below the threshold at which the precedence effect breaks down, so there is still a single localized image. They fall into a region where there are varying amounts of "image shift" - the image is either perceived to move slightly or to be stretched slightly in the direction of the reflection. I, and others, spent hours in anechoic chamber simulations of direct and reflected sounds and can confidently state that the effects, while audible in direct A vs. B comparisons, are rather subtle. Was it ever unpleasant? No, the apparent size and/or location of the sound image was just slightly changed. The effect was smaller than tilting the head a small distance left or right of precise stereo center. The dramatic change happened when the precedence effect broke down and two images were perceived – that was a problem. The strength and spectrum of any reflection depends on the strength and spectrum of the sound radiated in that specific direction by the loudspeaker, and by the frequency-dependent acoustical performance of the reflecting surface. If you look at (a) in the preceding diagram, the adjacent side-wall reflection is the sound radiated at close to 90° off axis from the loudspeaker. This is much attenuated in most loudspeakers, and is motivation to angle the loudspeakers to face the listener.

Are We Eliminating Reflections or Simply Modifying Them?

There is a fundamental incompatibility between normal specifications of acoustical materials and their behavior in normal listening rooms. The standardized specification of absorption, the random-incidence absorption coefficient, is measured in a reverberation chamber that is deliberately very hard and reflective so as to create a diffuse sound field. It all began with Sabine, who created the notion of reverberation time, and the specification is intended to help acousticians design relatively large, reverberant, performance spaces having significantly diffuse sound fields. I actually designed such a room, a recording studio that held a 75-piece chamber orchestra.

Figure 7.6: RI vs Angular Absorption

But, normal listening rooms and control rooms are not diffuse sound fields. With reverberation times of the order of 0.2 to 0.4 s what we hear from loudspeakers is the direct sound, a few early reflections, and not much else. Classic diffuse reverberation does not exist, nor would we want it to.

In (a) this illustration (based on Figures 21.8 and 21.9 in my book) shows the random-incidence absorption coefficient of 2-inch, 6 pcf rigid fiberglass board. A naïve interpretation of this tells is that this material will absorb everything above about 500 Hz and a sound incident upon it would be completely attenuated above that frequency.

But that is for a diffuse sound field in a reverberation chamber. In a listening room, we need to know what happens to a sound from a loudspeaker that reflects from this material when it is placed at the point of reflection, as seen in the preceding illustration. In that situation we are interested in knowing how much of a sound that arrives from, say, 45° is reflected towards the listener. Illustration (b) shows this as an attenuation of the sound. If we use the data in (a) we would possibly expect something like the dashed curve that falls rapidly above about 100 Hz, essentially eliminating everything above 400-500 Hz. What we actually get is something quite different because resistive absorbing materials change their absorption properties as a function of the angle of incidence. They also change when the fibrous absorber is covered with an acoustical fabric – the one shown here is probably the most widely used fabric in the industry (Guilford of Maine FR701), and it is clearly not acoustically transparent at very high frequencies. This absorber has not eliminated anything, it has simply turned the treble down – the off-axis performance of the loudspeaker has been “redesigned” especially if it is a well-designed wide-dispersion system. There are other published examples of real-world installations where the attenuation has a similar shape and an even lower tilt. These devices change the sound from the loudspeaker without eliminating the reflection; for that they need to be much thicker.

In reality we hear the combination of the off-axis performance of the loudspeaker and the frequency-dependent absorption of the material at the reflection point. For a lot of conventional forward-firing loudspeakers, the 90° off-axis radiation and therefore the reflection are substantially attenuated above about 500 Hz. But the remainder of the reflected sound is still audible. Figure 6.18 in my book shows that eliminating everything above 500 Hz did not substantially change the audibility of a reflection, although it changed the sound quality of the reflection, and the apparent level as revealed by a broadband ETC.

This is information that is almost never known by people offering opinions about the audible effects of these reflections, yet it is known to be critical to the listening experience. If the spectra of the direct and reflected sounds are significantly different, the reflections are likely to be more noticeable, from subtle timbral effects up to a premature breakdown of the precedence effect, at which point listeners may be aware of two simultaneous sound images, one located at the loudspeaker and one located at the point of reflection. This is obviously not good. Over the years this is likely a factor in listeners rating loudspeakers with uniform directivity more highly than those with uneven directivity. Wide dispersion seems to be good, but especially if it is uniform with frequency and the spectra of the reflections is not substantially altered. Hundreds of loudspeakers auditioned by hundreds of listeners in double-blind evaluations have demonstrated this; it is monotonously predictable.

Humans evolved while listening in reflective spaces, and are comfortable listening in them. In fact, it is now widely recognized that we perceptually "stream" the sound of the room as separate from the sound of the sources - that is what happens in live performances. A Steinway is a Steinway; only the hall changes. Performance halls generally don’t have room mode problems because they are so large. The parallel situation in sound reproduction is that a good loudspeaker is a good loudspeaker, and its virtues are appreciated in a wide variety of rooms – except for the differences in the bass region. Section 11.3.1 in my book describes an elaborate experiment in which three excellent loudspeakers were evaluated in four different rooms. Listeners sat down in one room, went through a double-blind evaluation of the loudspeakers, with randomized location changes with repeats, and recorded their preference scores. They then moved to a different room and did it all again, and again in two more rooms. In the end, the statistical result was that the factor “loudspeaker” was highly significant: p = 0.05, and “room” was not a significant factor. The rooms were very different, but the listeners appear to have adapted to their individual characters, made the appropriate allowances, and proceeded to evaluate the loudspeakers – which they did with remarkable consistency.

There is much more to this aspect of the story, and now room adaptation itself is a serious topic of investigation. Recent papers indicate that something as basic as speech intelligibility significantly improves after a listener has had an opportunity to become familiar with the room. Most remarkable, is that this adaptation can occur in a matter of seconds. As I said, humans, because of binaural hearing, are well equipped to deal with reflective spaces. Two ears and a brain are much “smarter” than a microphone and analyzer.

Time Changes Things

But back to the main theme. The widespread belief that first reflections in listening rooms are bad originated in the recording industry, where many mixers felt that they were better able to do their jobs when they were in a strong direct sound field; reflections attenuated. The notion has been around for decades, and while it is widespread, it is not universal.

Early on, I recognized one possible contributing factor, based on the performance of some of the popular monitor loudspeakers. Most of them won no prizes for their on-axis performances, but off axis they were even worse; some truly awful. If these off-axis sounds were allowed to reach listeners, the sound quality could not be good. The UREI 811B (shown here) was a favorite around the time the “dead end” control room emerged. Here is proof of a purely electro-acoustical reason to fill the front half of a control room with fiberglass.  The 45°, 60° curves describe the potential laterally reflected sounds. There were other real and imagined justifications as well. In our double-blind listening tests at the time this loudspeaker had a strong “personality” and established a new low in sound quality ratings. One has to wonder what effect it had on mixes! 

As an illustration of how much loudspeaker technology has improved over the years, these data on the JBL Pro M2 indicate that whatever one’s opinions of loudspeaker/room interactions were in the era of the UREI, they cannot be the same in the era of the M2, and any similarly “neutral” loudspeaker.  Because it is desirable that the direct and reflected sounds resemble each other, the newer loudspeaker has an enormous advantage. Traditions need to be put into context, and some of them relegated to history.

JBL Pro M2 Spinorama Data - courtesy of Harman

But, in the audio industry traditions are durable, and some studios still advertise proudly that UREIs are available, along with the comparably bad sounding Auratones created to represent the crummy CRT TV-set loudspeakers of the time. The pro version of the Yamaha NS-10 replicated the basic sound of the Auratone, but with a bit more bass. Because bad sound cannot be “standardized” no single loudspeaker can “represent” it. Good, neutral, sound is the only goal that makes sense – but I digress.

What Do Listeners Prefer for Small Room Acoustics?

These are just the acoustical factors, and whatever truths they convey, there is the matter of perception – what do we hear and what do we prefer to hear? Over the years several people have asked the question. Here are some samples.

Figure 8.2 in my book shows some results from Kishinaga et al.(1979) in which acoustical engineers expressed a preference for reduced sidewall reflections for evaluating audio products, but normal sidewall reflections for "fully enjoying the music".  Kuhl and Plantz (1978) reported that audio professionals had a preference for a strong direct sound field for mixing, but for more reflections when listening at home.

My own investigation of the importance of loudspeaker directivity as revealed by the energy in lateral wall reflections (Toole, JAES 1985 and section 8.2.1 in my book) examined details of stereo soundstage and sound quality. The principal conclusion was that the recording technique is likely to be the prime determinant of directional and spatial impressions in stereo listening. The loudspeakers and the lateral reflections they generate are factors, but not consistent factors.  Differences in soundstage and imaging qualities were thoroughly interrogated and found to be similar in all configurations. The effects of early lateral reflections on sound quality were most revealed by monophonic (hard-panned) components in stereo recordings or simply listening to a single loudspeaker.

Michelson and Rubak (AES preprint 4472, 1997) examined the recording techniques that influenced perceptions of distance in stereo recordings. They did the subjective evaluations in an anechoic chamber and in an IEC 268-13 room (a typical domestic room) and found no differences attributable to the room. One would think that this is a dimension of auditory perception with a degree of sensitivity to extraneous reflections − apparently not.

Choisel (2005, PhD Thesis, U. of Aalborg) interrogated stereo listeners on precision of sound image location when sounds were delivered by a very-wide, 180°, dispersion loudspeaker (B&O Beolab 5) or a normal, narrower-dispersion loudspeaker (B & W 801N), when side walls were reflecting or absorbing. The conclusion was that the perceived directions of images were unchanged by any of the manipulations, even though they resulted in large measured changes in reflected energy.

Prompted in part by my book, staff and students in the McGill U. tonmeister program did an elaborate blind-listening study evaluating the ability of professional sound mixers to perform with randomized reflective, diffusive and absorbing side walls. In two different control-room situations, they found that the mixers adapted quickly to all of the sidewall variations and just got on with the job. Any changes in the mixes were minor, and not statistically significant. Individual preferences among the mixers included all surface treatments, reflective, diffusive and absorbing, about equally (King et al., JAES, 2012).

A recent study reported that mixers preferred attenuated lateral reflections, but mastering engineers preferred a more normally reflective room.  In both cases, results were different with different songs. (Tervq, et al. JAES, 2014). 

Siegfried Linkwitz, a significant contributor to audio, has exorcised the demons of stereo reproduction for years, and currently listens in a relatively reflective room fitted with multidirectional (dipole), or widely-dispersing monopole loudspeakers. Each is apparently able to deliver similarly good stereo sound and soundstages. He concludes that “Monitor and consumer loudspeakers must exhibit close to constant directivity at all frequencies in order to obtain optimal results for recording and rendering.”  (www.linkwitzlab.com, “Stereo recording and rendering 101”). Because this is not the normal circumstance, recordings are highly variable, and he concludes “The aural scene is ultimately limited by the recording.”

This is what I have called the “circle of confusion”. The result is that the most “perfect” loudspeaker in the most “perfect” room, will not always, or possibly ever, sound “perfect”.

In all of our loudspeaker evaluations, over several decades now, the highest scores have gone to those with the most uniform directivity, not the highest or any particular directivity. In fact one of the highest scoring loudspeakers for a period of time in the NRCC double-blind evaluations was an essentially omnidirectional bipole design, which would generate maximum reflections from all vertical surfaces. This makes sense in that the precedence effect would remain intact because there is a spectral similarity between the direct and reflected sounds. That was a learning experience.

The Phantom Center Channel Superiority Myth

A possible factor in the acceptance of some reflected energy may be related to the interaural crosstalk that audibly distorts the phantom center image in stereo.  When in the stereo seat, in the absence of reflections there is a huge 2 kHz dip - it is not hard to hear when one is in a dominantly direct sound field. As shown in Figure 9.7 (below) it is somewhat alleviated by reflected sounds - even improving speech intelligibility.  Remember that normally this is the sound of the featured artist. There can be no better argument for a center channel. It could have been done years ago, but, no, we must be content with two-channel mediocrity. No amount of “High Rez” audio can fix this. The bandwidth should be devoted to delivering more channels (that is my opinion), and these days we can have both. What are we waiting for?

Figure 9.7: Real vs Phantom Center Channel Frequency Response Comparison

A second factor affecting how listeners respond to reflected sounds is hearing performance. We all know that exposure to loud sounds causes hearing loss. Normally we think of not being able to hear small sounds when thresholds are elevated, and that is true. What is insidious is that we also lose some ability to differentiate between closely spaced sounds in the frequency domain; even perceptions of consonance and dissonance change. We also perform less well in complex sound fields: multiple sounds simultaneously occurring in space – the cocktail party effect. That spills into a reduced ability to separate room sounds from the sounds of voices, instruments and loudspeakers reproducing voices and instruments. This latter disability has been found in persons with quite normal audiometric thresholds and therefore has acquired the name “hidden hearing loss”. This is not a nice situation.

Hearing Capacity and Its Influence on Sound Field Preferences

So, what does this mean? It means that for those with deteriorated hearing everything is clearer, easier to comprehend and analyze when in a simple sound field. Attenuating reflected sounds would be a good thing. It is well established that hearing loss is an occupational hazard among recording engineers and musicians, which means that mixers may have a physiological reason to prefer rooms with fewer reflections. The popularity of near-field monitors on the meter bridge of the console is another clue that a strong direct sound field is helping mixers perform their tasks.

Persons with normal hearing hear things differently; they hear more sounds and more spectral and spatial complexity in sounds.  They therefore have legitimate reasons to have distinctive opinions about the interactions of loudspeakers and rooms. In our early listening tests it was the highly variable sound quality ratings offered by recording engineers that led me to recognize the connection between hearing loss and judgment capability (Section 17.4 in my book and JAES, Jan/Feb, 1985). Subsequently, all listeners had to have audiometric tests and pass a training program before participating. I was an excellent listener for many years, but at a certain point it was clear that the standard deviations on my judgments were creeping up, and, frankly, I found it more difficult to form clear-cut opinions. In our listening tests records are kept on listener performance, as well as loudspeaker performance, and my age and degraded hearing were showing. I retired from the listening population. Figure 19.4 in my book shows my hearing thresholds from age 30 to 60, and thresholds belonging to several professional recording engineers who obviously were not hearing all of the sounds they were mixing. I still very much enjoy movies and music, and I have opinions, but they remain my own, not for public consumption.

The argument that recording engineers are greatly experienced professionals, and therefore can compensate for these unfortunate afflictions, does not hold water. Electron microscope photographs of the innards of damaged ears show large areas of dead and crippled hair cells and synapses between the hair cells and the cochlear nerve that communicates to the auditory cortex. Some information about the sound field simply does not reach the brain. The “microphone” is broken.

And Now, What “Toole” Really Believes

By now I hope that readers have concluded that the matter of early reflections in rooms is not a simple one. There is no single “right” way to do things. About 37 years ago, when I was setting up the NRC listening room, I ran a drapery track down the front portion of the side walls and across the wall of the room behind the loudspeakers, hanging 4-foot sections of densely-folded heavy drapes. The track was about 6 inches from the wall for good broadband absorption. These could be moved around, and in the case of the sidewall reflections, we quickly found that things sounded better if they were pushed back for more "spacious" classical music, and pulled out for "in your face" rock/pop stuff. I knew a couple of stereo enthusiasts who copied the idea at home. I concluded that, in terms of loudspeaker/room combinations, one size does not fit all. I believe it still to be true, but now we know a lot more about the factors that influence our opinions. Given the enormous variations in recordings, the absence of useful information on most loudspeakers, and the uncertainties of loudspeaker/room interactions, there can be no absolutely predictable outcomes. However, as has been found in many situations, human adaptation seems to save the day.  We tend to get used to what we have, up to a point.

The Quest for the "Perfect" Room

So, is there a “perfect” room? No. It all depends on your expectations, why you are listening (business or pleasure), what program you are listening to, and the condition of the ears you are listening through. In Figure 22.3 at the end of my book presents suggestions for room treatment based on guidance from research. The portions of side walls responsible for first lateral reflections are specified as “optional areas: absorb, diffuse, reflect.” It is a decision to be made by the customer and/or the installer. People who assert that I am in favor of lateral reflections obviously did not finish reading my book. In fact, as chance would have it, I do not have first lateral reflections in my current listening room – one wall is missing, and the opposite one is heavily draped for symmetry. It sounds just fine because the loudspeakers are superb. For multichannel movies and music, including upmixed stereo, early reflections are swamped by the important, and much louder, spatial cues in the surround channels.

For optimum stereo listening if your music tastes are as eclectic as mine, one really needs adjustable acoustics and, possibly, variable-directivity loudspeakers, but we know that won’t happen. We also need old-fashioned bass and treble tone controls to compensate for the unfortunate spectral balances in some recordings created by poor judgment (perhaps) and poor monitor loudspeakers and/or “room EQ” (probably) in control rooms and home studios.  Figure 2.4 in my book shows the enormous variations in bass response in numerous recording control rooms employing essentially the same loudspeakers. Recordings coming from these facilities cannot have the same spectral balance. So, as much as we might like to think that getting the playback system optimized may guarantee satisfaction, it is simply not possible. The following figure illustrates the problem – it is called the “circle of confusion”, and it exists because there are no standards in recording studios relating to loudspeakers or acoustics. Consequently, recordings are variable and it is left to the ingenuity of consumers to make adjustments, or to their tolerance to put up with the variations. The sad thing is that we are not listening to the “art” as it was created. If we don’t like something we hear, we have no way to identify where the problem originated, and it could very well be at the source. Pity.