Three Acoustical Issues that Room Correction Can’t Correct
Over the last 20 years, the increasing availability of cheap and powerful digital signal processing (DSP) hardware has enabled many audio companies to introduce “digital room correction” (sometimes abbreviated to DRC) devices. The first company to release a product in this field was SigTech, who were followed by a number of other companies – most notably TacT, who can probably be credited with introducing room correction to the two-channel audiophile market. In the last few years products like the DEQX have gone beyond mere room correction into full speaker management, incorporating driver correction and digital crossovers.
Despite these trends, much confusion still exists about what a room correction product does, what problems it can (and cannot) solve and therefore its place in a modern high-quality sound reproduction system. This article introduces three fundamental acoustical issues that exist in most listening rooms and home theaters that room correction cannot – for want of a better word -correct. These acoustical issues can only be addressed through good design, appropriate use of acoustic treatment, and appropriate system setup techniques (e.g. speaker placement).
Issue 1: Speaker Boundary Interference (SBIR)
Speaker boundary interference response (SBIR for short) is a little-known and poorly-understood issue that is responsible for deep dips or suck-outs in bass response below the transition frequency. SBIR is caused by the destructive interaction of the direct sound wave from the speaker and the reflected, indirect sound wave from a nearby boundary. If the difference between the direct and indirect path lengths (where path length is the distance the sound has to travel) is equal to half a wavelength, then the two sound waves will combine destructively and a notch in the frequency response will occur. The frequency that interference occurs at can be calculated through application of the wavelength formula as follows: cancellation frequency = speed of sound / (2 * path length difference). All of the boundaries in a small room can cause interference and audible suck-outs – the ceiling, front wall, back wall, side walls and floor.
A frequency response chart showing evidence of SBIR at 79Hz. A good measurement would not show any deep nulls. Note that measurements of frequency response in the bass region need to be taken with high resolution (here 1/24th octave smoothing is used).
Using a positive correction filter (for example, a 6 dB boost at 79 Hz) to attempt to remove this cancellation will not be effective, since the increased strength of the direct wave will be met in turn by the same increase in strength of the indirect wave. Reducing the magnitude of the cancellation depth can only be addressed through use of absorption at the reflection point on the boundary in question, which will reduce the magnitude of the indirect wavefront and therefore the amount of cancellation. If the magnitude of the indirect wavefront can be reduced by 50% then the cancellation expected at the null will also be reduced by 50%. The frequency that SBIR occurs at can also be varied by movement of the speaker and listening positions. Crossing over a speaker system to a subwoofer at 80 Hz, positioning the mains more than four feet from any boundary and placing the subs in corners will effectively remove SBIR effects, since there will be no difference in the direct and indirect path lengths for the subs, and the cancellation frequency for the mains will be below the crossover to the sub.
Issue 2: Strong Early Reflections
When we listen to music in a small room what we hear is a combination of the direct sound from the speaker and the multiplicity of reflections from the surfaces of our room. As Benade (From Instrument to Ear in a Room, Journal of the Audio Engineering Society, 1985) states:
“The auditory system combines the information contained a set of reduplicated sound sequences (i.e. the direct sound and its reflections) and hears them as if they were a single entity. The singly perceived composite entity represents the accumulated information about the acoustical features (tone color, articulation, etc) shared by the set of signals. It is heard as though all the later arrivals were piled upon the first one without any delay.”
In small rooms the contribution of reflections to the sound we hear can be as much as 60%. What we hear from a sound quality perspective in a small room is therefore determined as much by the reflected sound as the direct sound. The reflected sounds arriving at the ear are known to cause perceived changes in tonal color and increase the width and spaciousness of the soundstage. Because of these effects, the generally accepted target amongst acousticians is that all reflections be 10 dB lower in magnitude relative to the direct sound. Note that a sound at minus 10 dB is perceived as being around half as loud.
This diagram shows energy against time. Evidence of strong early reflections can clearly be seen where the plot crosses the -10dB line - there are a number of these prior to 13ms. A good measurement would show a plot that did not cross the -10db line after the direct sound at 0ms.
It can be stated categorically that a room correction product cannot reduce the percentage of sound reflected from a boundary. If a correction filter of -3 dB is applied, then it is true that the level of the direct and reflected sound will both be reduced by -3 dB, but the important thing is that the ratio of direct to reflected sound will still be the same. The level of reflections can only be controlled through appropriate use of absorption or diffusion or by selecting speakers designed to control directivity (i.e. waveguides, dipoles, horns). Some examples of these speakers are those from the companies Emerald Physics, GedLee and Gradient, which are all designed to control dispersion and reduce the amount of sound energy reflecting from adjacent boundaries.
Issue 3: Long decay times
Decay time is more widely termed reverberation time and is typically defined by the “RT60” measurement, or the time taken for the sound level to decay by 60 dB. From a purely scientific perspective the use of RT60 in small rooms is not valid since the science and reasoning is based on the assumption of a “statistically diffuse” soundfield, which is not a characteristic of a small room. From a practical perspective, however, RT60 is a good indicator of whether reverberation time is in the ballpark for high-quality sound reproduction. An experienced acoustic consultant or sound engineer will be able to detect an overly live room by merely speaking or clapping their hands and listening to the decay of the sound field. The target RT60 is different for two channel audio and home theater. Two channel audio needs a room with longer decay times to create the illusion of envelopment, since there are no surround speakers to assist in this regard. Excessive reverberation, even when obvious echoes cannot be detected, audibly reduces a system’s retrieval of critical acoustic cues such as the decay character of a recording studio. In addition low-level detail, micro-dynamics and tonal color all suffer if sound takes too long to decay within a room.
This is a decay time graph, where the vertical axis represents decibels and the horizontal axis represents time. RT60 is calculated based on an extrapolation of part of this graph, typically the part from -5 dB to -35 dB. The smooth black line, if extended to -60 dB would read about 1.1s on the time axis, which is the RT60 of this room. This is way too high for high quality sound reproduction - the reverberation time should be around 0.4s for a two channel system, slightly less for a home theater.
The decay time of the room can only be brought to the level needed for high quality sound reproduction through appropriate decorative choices, furnishings and the use of acoustic treatment such as absorbers. As Art Noxon, founder of Acoustic Sciences Corporation states “If you have no acoustic treatments in the room and you have a DSP processor, what happens to the articulation in the room? You’re still injecting energy into the room. You still have reverberation time. You still have the lack of intelligibility that you had before…DSP doesn’t address the decay rate factor of rooms, and neither does equalization. There is no electronic sound absorber for sale.” (TAS Roundtable, Room Acoustics: Audio’s Final Frontier, The Absolute Sound, 2004).
Conclusion
I hope you have found this article educational and that you now have a good understanding of the three acoustical problems a room correction product cannot fix. Room correction is not a magical cure-all that can rid us of the sound quality ills given to us by our rooms. It cannot solve phase interaction issues such as speaker boundary interference that cause audible bass suck-outs. Nor can it solve long reverberation times and strong early reflections that degrade imaging, sound staging and clarity. These issues can only be solved by good acoustic design, treatment and system setup techniques.
What do you think about the role of room correction in a modern high quality sound reproduction system? Did you realize that room correction can’t fix all the acoustical issues of our listening rooms?
Nyal Mellor, Acoustic Frontiers
Very informative, Nyal. Some listening rooms aren’t entirely enclosed, of course. In the end of my room where my speakers reside reverberation can clearly be heard with the clapping of hands (so called live end). In the end of the room where I sit the wall behind has an open passageway and an ascending staircase with an open column banister across most the wall to the ceiling. That just leaves the one corner opposite the opening to the room which is home to a large open backed record case (especially yours from IKEA). Reverberation is practically non existent here (so called dead end). I’ve read this is ideal. Not MY room, the live end/dead end thing. Would you agree? My speakers are about 4X their cabinet depth from the wall behind them and about 3X their cabinet width from the walls beside them. Thanks.
Very nice article Nyal. Very succint and to the point in explaining the problems and why they cannot be addressed with correction.
Thanks Brian for the kind words!
Jim – your speakers are pretty close to the sidewalls; normally I recommend at least 4ft away from the sidewalls or if they are closer then definitely experimenting with acoustic treatment. Depending on your personal preference for soundstage focus vs spaciousness then absorption (better focus) or combined absorber / diffusor panels (better spaciousness) work best here.
Glad to hear your recommendation about the distance from side walls. Mine are 43.5″ from the tweeter centers. I might have brought them further into the room if it weren’t for the fact my listening room is but 147″ wide (12.3′). Therefore my speaks are 60″ tweeter to tweeter apart. The tendency of most people, I think, would be to separate them further apart, but sitting at the apex of an equilateral triangle with my speakers there is the sensation of a wide and deep stage invariably larger that my room size. Seems to have fooled the walls. The presence of instruments is also very realistic. I use single acoustic panels along the side walls just in front of the plane of the speakers. Being conscious of room boundaries and their effect (reflections; decay times; etc) and making that part of the decision process when shopping for speakers has been good advice for me. After all, we don’t listen to speakers do we? We listen to what they portray in the room.
Any opinion about what is to be gained, or not, from the live end/dead end concept?
Hi Jim
Live end dead end was a concept that came out of the studio world. It was not a treatment scheme designed for home audio reproduction but for the control room. I don’t think it has any particular relevance to us as the priorities in a home environment are much different from those in a control room. Even the speakers we use are designed differently from those used in control rooms – off axis response is very important to the quality of home audio reproduction since we listen in reflective spaces whereas if I am a designing a speaker for a control room with absorption for all off axis sounds then that parameter is much less important than say maximum SPL.
Nyal
I am the founder of Juice Hifi and the creator of Audiolense, a computer based solution for sound correction and digital crossover. I hope my comments to this article is welcome here.
Speaker Boundary Interference: The frequency plot in the article is very representative of problems we often see amongst our customers. It seems like very few speaker manufacturers are successful in getting the deep bass and the lower midrange right at the same time. But this is a situation where digital sound correction can do a really good job.
If you boost a sharp depression, like the one around 80 Hz in the graph, with 6 dB, you will meassure and hear 6 dB more output of direct sound and reflections combined. It doesn’t matter that the caused by out of phase reflection doesn’t make a difference. As long as there is sound there it can be amplified.
This is a speaker with full range output and it will likely handle a lift of 15 dB around 80 Hz with ease if that’s what the doctor orders. A good correction will not eliminate the 80 Hz dip completely. The wide depression between 80 and 160 Hz will basically be gone, but a thin crack at 80 Hz will most likely remain.
The biggest benefit of using DSP is usually to get the overall frequency balance right. But there are a few pitfalls and limitations involved:
1) The freauency response – such as the one we see in the frequency plot is the stationary response of the system. If you play a 80 Hz note for a long period, it will settle on 68 dB output. But the signal may be much stronger before it settles. This is often the case. The dips aren’t relly as deep as the response plot shows. If you correct the dips 100% you end up with temporary peaks and get the hollow sound that everybody talks about. This is a general problem that can lead to overcorrection unless the measurement is properly analyzed.
2) A lot of people treat EQ and FIR based correction as if they were basically the same animal. Correcting such deviations with precision, using traditional EQ is however close to impossible. You get too little here and too much there, and possible a hollow sound because of boosting a few frequencies that are already strong. You need good frequency resolution and a flexible approach to make a mirror image of the response you wish to correct. A precise FIR based correction is what you need to obtain a really good result.
3) Correcting deep dips and high peaks and sharp wrinkles in general will require a filter that has more ringing than beneficial. You want to keep the filter as short and as simple as possible. So you get the best result by reducing the peaks and dips significantly rather than eliminating them completely. The best DSP solutions does this by default.
Strong early reflections: The strong early reflections will blur the stereo image and distort the timbre. A correction doesn’t remove the reflections, but it does restore the stereo image and the timbre very well. Strong early reflections will sometimes also have a negative impact on the midrange clarity even after the timbre is corrected. This lack of clarity is on my list of uncorrectable items.
Long decay: I agree with what is stated in the article. Too long decay tends to mask the ambience on the recording and other low level cues that are really important for a state of the art presentation of the music event. DSP falls short in this department from around 100 Hz and up. You can tighten up the bass but you can’t tighten up the decay.
A lot of the ongoing discussions of acoustic treatment vs dsp is limited by too much black or white thinking. A correction that is well done will always sound better than no correction. If you want to achieve the best sound possible you need good acoustics and DSP. And the icing on the cace is that a good time domain correction can compliment acoustic treatment, bringing the waterfall under control in the two bottom octaves.
I find the label “room correction” somewhat misleading. It is the mother of false claims and the father of straw men. What you can do with DSP is to optimize how the speakers interacts with the listener(s) in a given room. Nothing more and nothing less. The positive impact of Audiolense is usually dramatic when the speaker plus room behaves as the measurement in this article shows. The improvements are usually more subtle, but still significant in a well treated environment such as a monitoring studio.
Be aware that the differences between various dsp offerings can be massive. There are no unifying standards for how to interpret a measurement or how to make a correction. Two different DSP solutions are likely to perform different even though they seem to use the same principles.
Best Regards,
Bernt Ronningsbakk
Juice Hifi
Bernt
Very good points and I generally agree with everything you have said. For reference I am not anti room correction. I use it in my personal system as a compliment to traditional acoustic treatments. I would write some small clarifications on your points but they would be quite theoretical and maybe not so interesting for people who read this zine…
Of course the article was intentionally written to try and be black and white. I wanted to provide a simple to the point response to the marketing bs from the room correction companies who say that their magic box can fix all the acoustical issues of our living spaces. I am convinced that someone at these companies has a good understanding of what can and cannot be fixed but that it gets lost or glossed over as it goes through the marketing department. It is refreshing to hear from someone in this domain who openly acknowledges that room correction can’t fix everything.
For those with further interest in the topic I would direct them to the room correction series on my website at http://www.acousticfrontiers.com/whats-new/2010/7/1/room-correction-a-primer.html and http://www.acousticfrontiers.com/whats-new/category/room-correction.
Nyal,
I’m glad to see that my comments are well taken.
I hope I didn’t underplay the DSP card. The improvement our customers experience is usually very significant.
It seems to me that the solutions you have investigated doesn’t fully represent the offerings that are available. We are among a few companies that offers true time domain correction. It certainly doesn’t “fix the room” but it attenuates parts of the time domain distortions induced by the room in a very fundamental way.
The scope of the time domain correction in Audiolense is frequency dependant. You can correct a really long time window in the deepest bass. The speaker will then basically operate as an active bass absorber and speaker combined. A typical window is 0.25-0.5 seconds @ 20 Hz. In the other end of the spectrum we are limited to address microseconds; basically correcting the timing, phase and shape of the first arrival in the treble, perhaps with the earliest cabinet diffractions included.
It is highly variable how much time domain correction there is any point in doing. Sometimes it just sounds the same when you add more, sometimes there is a break-even point that varies from room to room. The time domain correction is very sensitive to measurement artifacts and measurement quality and also depending on good overall performance from the hardware (low distortion and high linearity). I don’t like to see a lot of grit and noise before the main spike of the impulse response when the user is aiming for a time domain correction.
I frequently use head phones for evaluation. I listen to the pure version, to a version convolved with various room + speaker responses and to versions convolved with the room/speaker response after correction. This has become an integral part of ongoing development but is also a valuable tool when customers need support. It enables me to hear something similar to what they are hearing. The coloration from room and speakers are usually very audible, even in rooms with studio grade acoustic treatment. The good news is; even with a moderate correction – as soon as you start to follow the music it is very easy to hear that the linear distortion (level & time errors) has been dramatically reduced in the corrected version. The room sounds the same but the speakers and the music sound clearly better, so to speak.
When I work on improving the correction algorithms I can apply almost insane quantities of time domain correction and get away with it. It works as a sort of magnifying glass. Then it is easy to hear that the corrected version has significantly less room coloration added than the uncorrected. This proves to me anyway that DSP has some capabilities towards negotiating room effects in a very fundamental way. In a real setup you have to settle for less time domain correction or you will get audible correction artifacts. The time domain correction doesn’t make the room disappear acoustically. But it usually changes the perceived sound quality towards a few dB more direct sound, and this is also measurable improvement.
The biggest win in using DSP is in any case always related to the frequency domain correction. That may not be as sexy on paper as time domain correction, but the frequency response of the direct sound and reflections combined has impact on a huge number of sonic qualities. Some of them are usually not associated with the frequency domain, such as sound stage and resolution. When I discuss DSP with acousticians, they usually emphasize the time domain behavior and I usually find myself defending the significance of the frequency response in the listening seat. In a way we are both right, since the biggest frequency domain problems is caused by (delayed) reflections.
The quality of various frequency domain corrections are highly variable, and you can almost talk about a tradition where mediocre frequency correction is established as the norm. I regard that as the main reason why DSP is underestimated by quite a few audiophiles. For some mysterious reason a lot of audiophiles seem to believe that negotiating baffle step issues, driver issues and Allison effect like issues with high power resistors, inductors and capacitors inside the speaker is as good as it gets. And that doing the same task in the digital domain with far greater resolution and flexibility, and without the heat buildup and distortion associated with passive crossover components is a less puristic solution. Go figure.
You need FIR filters with a long duration to achieve proper correction in the bass with good enough frequency resolution. Not all available solutions provides that. But the most important part is perhaps that the DSP unit filters and analyses the measurement in a psychoacoustically adequate way: A correspondence between how it looks on the graph and how it sound. And this is the hardest part. In theory you can dial in the sound by ear as long as you have enough frequency resolution, but in practice it is almost impossible. For the same reason you also need a calibrated microphone to dial in the system properly. I also noted that you assessed the ability to manually adjust the target in various offerings and I can’t overemphasize the significance of a custom target.
Acoustic treatment and DSP is a very potent combination. Acoustic treatment can negotiate all issues except the deepest bass very well. You can basically choose the decay pattern in the room by design. DSP can compliment and perhaps even do the heavy lifting in the lowest octaves time domain wise where the size of effective diffusers and absorbers takes on abnormous proportions, but dsp can only scratch the surface of the time domain for most of the frequency band.
When the time domain is cleaned by room conditioning the frequency domain becomes much smoother. The roller coaster like frequency response that was before is now reduced to a bumpy road. Part of the bumps will be due to a speaker that was a bit bumpy out of the factory and parts will be remaining of the before acoustic problems. DSP can then be used to almost completely eliminate the frequency bumps. And to tighten up the time domain behavior further. And take the system up to a sound quality level that is unattainable without both acoustic conditioning and DSP. A combination of acoustic treatment and DSP is a very potent one. That’s why I like to discuss these matters with people who covers the middle ground between DSP and acoustic conditioning.
Nayal what is your opinion in using a parametric eq in JRiver media center to flatten a frequencie range from around 600 to 1300? I am flat from 20 to 600 with a 8db dip At 600,then a 6db peak from 900 to 1300. My room is well treated and I have been able to measure great bass response with the use of three subs in the room. My room size is 13×29 I have Magenpan 1.7 set up on the short wall pulled out about 7 feet from the back wall, listening chair 10 feet from speakers.
I use the ominimic v2 for my measurements. Would you reconmaned using a parametric to flatten the above mention problem area or try to fix the problem in the room? Thanks, Sam