Speaker extremes

August 24, 2016
 by Paul McGowan

Tomorrow I want to move on to another way ofย dividing frequencies in speakers, the amplified electronic crossover. But first, let's wrap passive speakers up with a look at a very expensive one.

Wilson Audio has always been among my favorites. Sure, every loudspeaker has its fans and detractors, but ever since the first day I met Dave Wilson and he encouraged me to sit in front of a pair of Watt Puppys, I have been a fan. They disappear nicely, always sound musical, leave me wanting for little.

At $200,000 the pair, the Wilson Alexandria XLF ain't for the faint of heart. Imposing, full range and musical, this loudspeaker is arguably one of the best in the world.

So, what's it look like under John Atkinson's measurement microscope? Here's a graph of its frequency response as averaged in reviewer Michael Fremer's listening room from a 2012 review in Stereophile.

Wilson Alexandria XLF

Not bad, right? Reasonably flat and in a room too. But reasonably flat isn't "flat". Each horizontal line on the graph represents 5dB, and while the average runs in a relative straight line, the particulars are all over the map.

And I am not singling out this fine loudspeaker.

It's one of the best, and that is the point.

Loudspeakers are like unpolished children. You don't want to look too closely.

A lot of forgiveness is what's needed.

Subscribe to Paul's Posts

95 comments on “Speaker extremes”

  1. That's not at all a good foundation for valid comparisons of most delicate turntables! ๐Ÿ™‚ And keep in mind: these curves showing too many peaks and dips are akready smoothed by the measuring algorithm. And it seems that Mr. Fremer doesn't have the possibility to offer symmetric conditions for his speakers - assuming that speakers in this price range measure identical in an anechoic chamber and Mr. Atkinson is a most diligent guy setting up his measuring equipment.

  2. Some wonder about so many different opinions regarding tonality of various components.

    I wonder about so few, due to the fact, that each voicer's/listener's/reviewer's speaker (and room) has very different peaks.
    Hard to imagine how components would sound if indeed checked with neutral accompanying components and really flat measuring speakers in a room with flat response...I assume, even the developers would wonder about the sound of their products in a truely ideal environment.

    1. The only perfectly flat environment is an anechoic chamber, and that sounds terrible. We need to feel the space around us by the sound reflections. The absence of sound coming from any direction indicates either complete emptiness, which is dangerous coming from the floor; a coffin like enclosure, which is a threat of suffocation; or something non-reflective blocking the sound, which is also a potential threat.

      1. Yeah, I had quite some listening sessions in a sound engineers optimized room. Great sound for me, especially in bass at the time generally, but he wanted some more reflection to get some life into the music.
        So what I meant is more like a room with "ideal" delay times (which are not =0 ms)

        ELK recently spoke of kind of "mastering standards" that exist and keep recordings quite similar usually. Thats ok with me. Additionally I just searched for the "standard monitoring speaker" in the "standard room", not necessarily for mastering but at least to voice "neutral" products ๐Ÿ˜‰

      2. But listening to an open air concert (unplugged) in a park (not a stadium) I hear no reflected sound. What I could hear additionally (but do not focus) is the ambient noise!

        1. You need to go to a concert in an isolated place with a bowl shaped hillside or natural amphitheater. Red Rocks comes to mind, concave back wall and good side wall reflections. I also enjoyed Telluride and Winter Park Festivals.

        2. What if I tell you open air is better for . Indoor concerts that echo is muddy to my brain. A moving target that fatigues me . A clean studio sound is best for me. . But I do get the point ona m odd man out .

          1. I was raised in Pine Forests and sailing on bays, so I am acclimated to both heavy and light diffuse reverb. On the water, sound travels long distances and reflects off waves so the acoustic change with wind strength.

            Composers have an intended acoustic, which is usually implicit. The most obviously explicit are Handel's "Water Music" and "Royal Fireworks Music" which were intended for an orchestra and Royal audience floating on the Thames, before the banks were over-built.

            I once heard the original orchestration - 9 Baroque Oboes, 9 Baroque Bassoons, 9 Sackbuts - which gave 20dB boost from the line source and rise in radiation resistance to be heard in an low echo environment and competing with gunpowder blasts.

            Unfortunately, it was in a Cathedral. I suppose a motorized barge on the Hudson in January was not a good alternative :^(

  3. Nothing wrong with this post; pity the name Fremer is mentioned.
    From what I've read and heard (Youtube) from him he's on my list of most suspicious audio reviewers.
    Give him a turntable and he starts to drool, a cd-player and he starts to vomit. So far for an open mind.
    Okay, I had to get that off my chest. Sorry for nagging. Off topic, totally misplaced, I know.
    ๐Ÿ˜€ And besides, I guess he couldn' t make a mess of this graph. It's not a review.
    Probably need to see a psychiatrist to get rid of this Fremer complex.

    1. Have you actually read Fremer on a regular basis? He really doesn't need me defending him, but I have read almost every column and review, I have been reading Stereophile since 1993. Yes, Fremer is pro vinyl, in fact if you like vinyl, you should be thanking him. He kept the light on when tin eared audiophiles were selling their record collections, replacing them with "perfect sound forever" CD players. Digital audio has come a long way from the early days. I am one of those people who waited years to buy my first CD player, and have every record I bought since 1969. Well, except one, a 45rpm I sold for $425.
      When analog vinyl needed a voice it was Fremer, who spoke the loudest.
      The only way to determine a reviewers worth, is by reading them for a long time. Are they consistent, do components they recommended sound good when actually get to hear them, preferably in your own system.
      Fremer reviews digital, and yes, he will say it still can't compare to his $200k reference, but he has given positive reviews, many times.
      As to Atkinson's testing, if flawed, it is still consistent, you can compare speakers tested the same way, and see the differences. I have an spl gauge, is it calibrated, no, but it will measure differences, that I can use, based on the first measurement.

      I am more suspect of the many online reviews that come from people I have no history with. That doesn't mean they aren't honest, or have bad ears. It means I have no point of reference.

      And posting on this newsletter because Fremer's name was mentioned, that's kind of lame.

      1. "Fremer reviews digital, and yes, he will say it still canโ€™t compare to his $200k reference..." As an absolute statement.
        That's where the trouble with Fremer starts for me.
        I don't like reviewers saying this or that IS better, instead of making clear it's just their opinion.
        For me a 200k digital system sounds better than an analog system.
        And I've heard some. Maybe you like analog. I don't.
        So Fremer and I wil never be on the same page.
        And if you can't stand other opinions, that's...erm well lame.

        1. My opinion is that I can't take anyone seriously who demagnetizes plastic phonograph records and then says they sound better that way. Atkinson, a glorified underequipped bench tech once said he heard a difference but never took the demagnetizer into his lab to measure if it had any effect at all. We're talking amateurs here folks.

      2. Both vinyl and tape best bad digital I agree there. But my best stuff is from to tape to dsd , played back well it beats vinyl and is a cheap simple way to reproduce our music over a life time. I am not sure vinyl is as great as most claim.
        Keep in mind it's multigenerational away and eq as well. Vinyl for all of its praises is far from pure , tape is much better . Has anyone hear heard a laquer of an album ypu know . It's amazingimg much better. Well that's tape of you get a second gen copy , even a third is still very good .

    2. Actually, I save my scorn for Andrew Quint. He gets a set of speakers, room EQs the hell out of them, and then says they sound good.

      I do not have a problem with room EQ but I don't see how EQing speakers to a "known flat response" makes it possible to compare them. They should all the same at that point...

      1. Since speakers and rooms are not minimum phase, normal EQ algorithms make the phase response worse when they correct frequency response. This was covered in J. Robert Ashley's speaker design course I took at the University of Colorado in 1976, and it was not news then.

        Further, frequency response anomalies in rooms are spatial resonances so correcting the steady state response makes the transient response worse.

        You can't flatten rooms accurately with state-of-the-art convolution technology like Room Perfect because the resonances go chaotic.

        The most expensive part of an accurate reproduction system is not the gear. It is a gut re-model of your room and the adjacent rooms in your house, replacing all the furniture, the architect, acoustic designer and room tuner, and replacing your music collection with proper recordings - which you will have to do yourself unless you can be happy with only Chesky, 2L, Lipinsky, Aliavox and similar labels.

        1. [@acuvox]
          Can you please tell me which loudspeaker with digital room correction was used in 1976?
          Otherwise comparing analog room correction and digital room correction is like comparing apples with pears.
          I've heard loudspeakers that sounded quite good without digital correction but even better with it. So if no better way is found to solve the problems and as far as we have no better tools to overcome the obstacles a digital solution for room correction is a good start.

          1. In 1976 there was no digital - except for a recording console I built that had digital buss assignment with a memory.

            I have heard digital correction improve sound over a limited range of seats, but there are a lot of limitations, the most obvious is it applies in one dimension (voltage) but sound waves have three. Rooms produce response that can vary >+-10dB over distances of three feet, so correcting the sweet spot can make the next chair worse. Speakers can have spatial nulls in response of similar magnitude. Besides, any variations that are not minimum phase can't be corrected in both the frequency domain and the time domain. Both room and speakers have non-minimum phase deviations.

            I am dealing with this using digital noise removal software. I make live recordings of concerts - over 600 by now - and they are always interrupted by some noise. Even in well isolated rooms the noise of the city intrudes occasionally, musicians make incidental noises and audiences can't sit still. Digital filtering allows these noises to be expunged in many cases, but eliminating them carries a heavy price in the phase of the remaining music, as does digital file compression.

            I am acclimated to phase accurate music. I listen to live music more than I listen to reproduction, and I only listen to my own live, unprocessed recordings because I can hear processing. Because of this I can only listen to noise-removed music for an hour or so before my hearing starts to scramble, likewise streaming music gives me a headache after an hour and I go phase deaf for a while, the same way that exposure to loud noise raises your hearing threshold.

            I don't know how much digital room correction causes "phase fatigue", but I am not going to pay for the experiment. Where digital does work is correcting the phase and frequency of digital speakers, like Bruno Putzey's work.

            1. [@acuvox]
              I understand quite well because I had to keep all this in mind when I built an ADC by my own using FPGA technology.
              Besides all you've said I still do not understand how a digitally corrected FR should damage the sound.
              Done the right way the advantages are more than convincing.

              1. There are several factors. The first is that any digital equalization requires delay and addition, so it is inherently conflating two different points in time. The same is true of analog equalization, it has to introduce time/phase shifts. Digital can use negative time, but it still has a tradeoff that says you can't change the frequency response without distorting time.

                Second, there is the reality that every digital recursion reduces resolution. When you multiply and add, you have to re-normalize and there is roundoff error. In audio, distortion compounds geometrically because distortion of distortion sounds worse than distortion of signal, and this is exacerbated by dynamic range. If you are listening to signal at -30dBFS and there is distortion at -70dBFS, you can hear the distortion because it is 1% of signal.

                When studio processes pile up dozens deep the digititus becomes palpable. This is worse for time quantization than for level quantization so processes that have to cover long time period like bass equalization, compression and especially digital reverb which has hundreds of recursions over seconds, it becomes quite unreal.

                1. [@acuvox]
                  Please keep fair!
                  What about the amount of phase shift in an analog network and the digital on the other side.
                  I think that your thoughts are like "green isn't green enough it has to be greeeeeeeeeeeeeeeeeeeen!"
                  BTW entropy rises every time information is given further and that is a function of time.

                  1. I mentioned the time distortion of analog equalization. I am against any equalization besides necessary conjugates EQ like RIAA and NAB. I use only acoustic EQ in my recording process, the inherent timbre shifting of musical instrument geometry, microphones and room acoustics.

                    I also have strict interpretation of Green. Since mining is inherently unsustainable, irreparable and toxic, no mined materials can be "green".

                    Sustainability is an absolute, 99% sustainable is NOT sustainable. It is limited to photosynthesized materials which are 100% renewable, 100% recycled and 100% bio-compatible.

        2. Great post. While I cannot say what I like but I do know when I hear it. I have yet to hear a room corrective process that doesn't make my head hurt. The curve in my room moves plenty . But I can play music at live volumes , loud and not get Schrills on most of the music.

  4. Not unpolished children, flawed engineering. Besides their ragged FR, here are some more important failures IMO;
    They don't sound the same from one room to another, no provisions engineered in for that
    They don't sound the same from one place to another place in the same room, no provisions engineered in for that either
    They don't sound the same from one listening position to another in the same place in the same room, poorly engineered for that too
    They don't sound the same from one system to another, flawed system engineering for that
    They don't produce the required sound fields according to a mathematical model taking into account the commonalities from one musical instrument to another, flawed understanding of sound fields and room acoustics.
    They cannot produce a stable image from one location in a room to another, two front channels are not enough, you need five or more and not electrically derived channels but actual recording channels.

    Look at Atkinson's Youtube presentation about how he measures speakers. It is pathetic. Usually he measures them on his backyard patio using FFT which he readily admits has serious flaws. In the case of Wilson Alexandria he measured them in Fremer's driveway because they were too heavy to move. TAS doesn't even make measurements but it hardly matters because what you hear doesn't correlate with what is measured, the measuring method is flawed. So are the conclusions drawn from the results.

    Look at the response to a step function for any speaker Atkinson measured. The results are pathetic. The tweeter moves first, then the woofer. They are neither time coherent nor phase coherent. In a sustained note they will produce irregular FR from one place to another ranging from +3db to total cancellation....wave mechanics 101 teaches that.

    You donโ€™t want to look too closely." That depends on what your goal is, buying or selling. The more you look and the more you know the worse they look.

    "At $200,000 the pair, the Wilson Alexandria XLF ainโ€™t for the faint of heart. Imposing, full range and musical, this loudspeaker is arguably one of the best in the world."

    What happened to yesterday's best in the world from YG? And they are less than half the price of Wilson Alexandria?

    Enjoy them all you like if you can afford them. They were made by audiophiles for audiophiles but they are all engineering flops if the criteria is high fidelity. Same ideas as 50 years ago only pushed to an absurd degree like Audio Note's TOTL speaker, a Snell type E that Snell originally sold for $1300 a pair pushed to sell for over a million a pair.

    1. Your comments imply there are better speakers that do better handling what you cite as Wilson faults. Can you list a few of these with perhaps some comments on them so we could search them out to audition ourselves? And no I am not a Wilson owner nor is it my 'dream' speaker.

      1. Only the ones I re-engineer for myself...and acuvox's. The others are all badly flawed. Lack of insight, lack of scientific knowledge about sound, lack of engineering skill, these tinkerers are doomed to fail no matter what they do.

        1. Soundminded. You leaving out the part where it's sounds good. I'll bet I do not like what you or audio does. And if I don't your answer will be my brain is broken. Two people here posted tracks one a novice one was a pro and someone I like to read about like you. The pro was worse then the ref guy. While both were flat , lifeless . No distortion just flat and not anything I like.
          My point is no one can make speakers or music for all. So what you perceive from a science perspective view as flawed can and is music to others. Wilson speakers are not on my list , point source while they disappear well are not for me. But that is also my view. Multi channel audio does sound better to me , but there is little to no staging . Now you 5 channel is a point that does interest me. Your post does very accurately list the issues well. But this is what we have and I have yet to hear your or his improved views maybe it's better.

    2. Every single limitation of sound that you just described in a room or venue, we hear every day, whether we're conversing [especially in groups], singing, watching TV or attending live music. So why should we expect perfection from our sound systems ?

      1. Not true. Live music even in your own home would sound different from anything you could get out of one of these machines. But then what do you expect from the kind of people who design and build this stuff. Even a novice could hear the difference between a real piano in your home and the sound of a recording from any of these. When you get to the sound of live venues for concerts, the disparity is far greater.

        1. Stay on topic Soundmind. Frequency response is Paul's topic today, not tonality, harmonics and everything else that distinguishes live music from reproduced. I have experienced such corrupted frequency response in live concerts as to make the experience unlistenable. I'll cite the Hollywood Bowl for example, which sound corruption permeates the live sound factors !

          1. FR is a far more complicated problem in time and space for sound fields than these clowns can conceive. FR is directly related to perceived musical tonality. Much of what you hear is reflected off the walls, ceiling, floors, and objects in your listening room. The speaker could be dead flat on axis but way off flat transfer function due to these reflections. For example, there are virtually no reflections of high frequencies from behind the speaker or to the sides. The speakers are not engineered to control those reflection or to adjust them independently from the direct field. In fact they deliberately beam high frequencies along a narrow angle. Look at the off axis response of the ribbon tweeter recently shown. It is horrendous. High end tweeters from companies like Scanspeak and Morel are no better. The idea that you can ignore the listening room is a fatal mistake and utterly naive. So is the idea that you can kill it off. It takes real skill to incorporate it as an integral part of a sound system and to use it to advantage. That is skill and insight these chumps don't have.

            1. I thought we agreed a few days ago that the problem with reproducing music that sounded "live" begins with the recording microphone. Now it seems all the blame is being put on the speakers and the room. It is all important, but we can only control certain parts of the total equation. I can choose my speakers and re-decorate my room, but they certainly will not let me into the recording studio.

        2. Gary of genisis audio did a video of his flagship playing witha live saxophone player . Even on my iPad the sound was large . So what you say is very true. Most speaker companies admit this while others pint to the recordings as well. I own some ultra highend hesadphones. Planners , stats and diaframe. While they have ther strong points. None do what my IRS v or even my smaller rs1b does.

      2. We learn to sort out the complexities of acoustic sound because there are consistent rules of physical sound generation and propagation. The problem is that recordings and speakers have very different rules of room interaction, and they are not consistent.

        Musicians learn to integrate room acoustics into their perception by listening to the same musical instrument designs in different rooms for hours a day since childhood. They know what a group of violins sound like in rehearsal halls, concert halls and their living rooms. The rooms cause frequency response variations, but musicians' ears sort that out by hearing each room reflection discretely, using it to triangulate the position of the instruments and the walls and accounting for the affect on frequency response. This requires very good phase differentiation.

        Audiophiles are phase deaf because EVERY KNOB IN A RECORDING STUDIO and every commercially available speaker mangles phase. Even "phase aligned' speakers mangle phase off axis. Headphones eliminate directional phase encoding, and so eliminate the spatial cues that musicians hear. Acoustic musicians have great difficulty playing with headphones because they are so data-reduced, and they have to re-learn to hear that way to become studio musicians.

        1. Maybe this is why so many recordings from the 50's are so cherished. Of course they were noise but so many were recorded live with no massive control console.

        2. Love your post. Your remark on headphones though I do not know why is true they are much cleaner then most speakers if the headphone and chain is good. What are my IRS v like ?
          Spec wise

          1. Line sources have several advantages:

            1) Much larger surface area - this means they can play at realistic volumes with minimal Doppler distortion, including full scale peaks. Because most speaker can't handle them, the peaks are engineered out of 99% of recordings, including Reference Recordings, Red Seal, etc. In orchestral recordings, standard practice is to back off far enough to get acoustic compression, higher reverb energy, high frequency attenuation and phase shifting of peaks to reduce the crest factor.

            2) Higher efficiency - doubling drivers adds 3dB in FREE acoustic energy because it increases the air coupling transfer. If the line source goes to the floor, the reflection doubles the energy again. This reduces amplifier power requirement, meaning you can afford a better amp, and also reduces power compression, and suspension non-linearities with respect to a single driver.

            3) The line source reduces the vertical room modes and evens the frequency response.

            4) The line source falls off 3dB with doubling of distance instead of 6dB for monopoint, so the sound level is more even throughout the room. This also reduces variations in frequency response from room modes by increasing direct sound at a distance.

  5. 10 - 12dB swings in the bass. I don't feel so bad about my own speakers' in-room response curves now! They have been designed for the room that they are in yet the frequency sweeps look like the Alps. Well, maybe more like the Shenandoah if I want to flatter myself. Adding filters to smooth out the response gets back to a previous comment... "dreaded trim pots". I just end up preferring the sound with minimal filtering.

    1. I had a 15 DB bass boost from 40 to -100. Moving them around reduced to plenty. Of course it reset my imaging so more moving the mid towers. It's total within about 10 DB including moving my seating severa feet. Any form of dsp hurt the texter of the sound

  6. Wouldn't it be most interesting to see the FR plot of the speakers and room our the products we bought were mainly voiced with/in?
    Then we might have an idea where the one or other perception comes from.
    As our own speakers/rooms are equally compromised, this certainly would just have a relevance in comparison to other components checked under the same conditions.

  7. This has really turned into an odd run of speaker bashing in the comments. "OMG, look how imperfect that graph is!! Therefore it cannot sound good!!" (quite aside from how the graph was derived)

    Despite all of the questionable engineering and bad speakers out there, the fact is, a fair number of speakers sound good, and not necessarily for lots of money. Actually money never seems to have been a good predictor of speaker performance.

    Unfortunately the things you have to do to them to correct the flaws - inherent in their designs, and in poor room response due to poor placement/bad rooms - may flatten out the response on a graph, but usually make them sound worse. I suspect evryone here has heard a device or speaker that tests nearly "flat from DC to Light" and yet fails to produce music.

    The irony is that our hearing is anything but flat and perfect. Even if your ears test well, we all have different shaped ears, heads and bodies, creating significantly different response curves that vary with direction (HRTF). And we have massively different response to level at different frequencies (see Fletcher-Munson).

    So don't shoot the messenger. ๐Ÿ™‚

      1. In this case, the messenger can't help but interpret, and most would not fault him for trying his best, or for doing so with a realistic or aesthetic mix of accuracy and artistry.

        There is a lot of talk of measurement and theory and so forth. When you create an actual viable product that satisfies your criteria, let us know, we would love to hear it!

    1. Biology USES the non-linearities to extract more information than a mathematically perfect system. Individuals have been documented up to 13 times better than the Fourier Uncertainty Principle, and also need more than ten times the bandwidth of satisfying the Nyquist criterion. The shape of human ears is to function as directional phase encoders so we can tell which direction sound and reflections of that sound are coming from.

      Every ear is different, so you can't encode music so it plays back accurately for more than one person at a time! Further, it takes thousands of hours of listeining in consistent low noise environments to develop the full capability of human hearing - and conventional speakers do not have the consistency to decode rooms using them.

      In fact, speakers for reproduction and PA systems have evolved to be locationally vague. If you can hear where the speaker is, the stereo or surround image collapses. There are no acoustic instruments which have the same characteristics - they are more sound anchors, with high localization parameters. This also affects music production, audible cues that make the speaker location detectable are erased from the recordings.

      Doing this the right way places much more stringent criteria on reproduction systems, because frequency response is a small fraction of the information content of directional phase discrimination. They have to replicate not only the direct sound, but also all of the reflections. Since it is impossible to do this with fixed channels, the entire audio paradigm is flawed.

  8. Since you are posting frequency response graphs of speakers, I've got a questions about them that I've wondered about for a long time and haven't seen an answer. Maybe someone can help me, I'm a newbie so this might be old stuff.

    1.) Is it safe to assume that the frequency response is constant at all decibel levels. For example. If two people in the same room with the same equipment listened to the speakers, would the frequency response that the person listening at an average of 78db be the same as the person who listened louder at 85db?

    2.) My understanding is that the frequency response graphs are made using test tones and each test tone is presented to the speaker discretely. What happens when the speaker is already playing other frequencies. So for example, is a speaker that is ruler flat say from 1k to 5k still ruler flat when it is also playing a 50hz tone. Can that be measured?

    1. The answer to question 1 is yes up to certain levels. If the speaker tried to get too loud, some of the dynamics possible may limit output - but for normal to reasonably loud listening levels, the answer would be yes.

      There are steady state tones that do what you suggest, but also pink noise testing - sounds like rushing air - that etss and plots the speaker while it's playing all frequencies at the same time. So, yes, a speaker can do more than one thing at a time, and successfully too!

      Great questions and thanks for asking.

    2. 1 Unless the drivers are driven so hard they the begin to suffer from compression or severe distortion, the frequency response will be independent of level. On the othe hand your subjective impression will change with volume level as the response of our hearing system foes chane with volume level.

      2 John measures using the DRA MLSSA system which does not use sine waves but presents all frequencies simultaneously.

  9. I've also got a question:

    assuming there's a tendency that 70% of customers have a quite common and typical room hight and maybe even width (leading to resonances) and assuming a speaker has a quite noticable peak/dip, deviating from the ideal in some area anyway and assuming a designer had the choice:

    wouldn't it make sence to influence this peak into an area where it rather helps than harms those more or less standard room measures?

    1. Great question, again, and yes. The skill of a speaker designer is to maximize deficiencies and minimize those weakness he cannot control. Take small two-way speakers as an example. The good ones, the ones that sound like they have bass, those (of course) don't have low bass, so they maximize performance and peaking at 60Hz so it sounds like it has low bass. It's a common technique and an effective one.

  10. This has become one of my arguments against the current audiophile madness. What do you get buying a $200k flagship vs a more entry level model from the same company?

    Generally you get the same tweeter. With the XLF mention by Paul, you get two mid-woofers. ( I'm not picking on Wilson, just look at any uber speaker line). Cabinet construction probably didn't change much and while they may go on and on about crossovers..... It's just a crossover. But you do get giant woofers in a giant box to feed your giant ego and the manufacturer/dealer gets a giant payday.

    Back in the day when a flagship was under $20,000 ( a topic for another post ) I too tried top shelf speakers. Sometimes I got lucky and had bass that seemed to sound wonderful. Sometimes my luck ran out and my bass was just terrible and un- fixable.

    Then a few years ago I read a subwoofer review from Jeff Fritz where he described using subs to smooth out the bass in his room. He then went on later I think, to tell how he has been able to get almost 100% of the performance of his giant speakers from a two point one system. Jonathan Valin has also made similar comments,especially when using an active crossover. Floyd Toole also devoted a chapter in his book about bass in small rooms. Basically his conclusion is that it is impossible to realiabily have good bass from large speakers and concluded that you need multiple subwoofers strategically placed around the room. Mr. Toole supplied impressive sound analysis to back up his claim.

    I have since then gone that route and have the best bass that I have ever personally had in my 40 odd years of working on this problem. I'll even say it's the best bass I've personally heard anywhere.
    Manufactures could do this but then there goes about half their profits. I would like to applaud Richard Vandersteen for doing basically this for many decades.

    I'm sure I'll be flamed by those you have giant speakers already. If you do, I hope your bass is better than what JA typically measures at the listening seat in -room at various writer's homes.

    1. Do your speakers also succesfully pass the LEDR-test? During the last 25 years I was not succesful in finding a speakers-room combination either demo-rooms of dealers or show-rooms of manufacturers where this test performed as claimed. Or is this test a pure kidding of audiophiles by a cynical sound engineer? ๐Ÿ™‚

      1. Sorry, I've never bought the test disk. What I do get depending on which speaker I have out at the moment is a pretty good illusion of Harry Belafonte and his band members walking around and singing on a stage in front of me. His live at Carnegie Hall disk or record, has sufficient information encoded to easily give the illusion of specific left right and front to back positions as well as which direction he is facing on a stage that appears to be about the right size. I'm sure a lot of people on this forum have rigs that would do this too. In a nod to Paul and his team, I never got this illusion from CD until I bought a DSD DAC. Point being is that this level of image precision is not always available from every piece of gear.

        1. I still have an original pressing of "Belafonte Live" that was one of the first records my father purchased. True audiophile recording, great music, and proof that live is better than studio. It was his biggest selling album, by far.

          Human hearing can discriminate TIME to better than 3 microseconds. This has been documented over history from mechanical apparatus of von Bekesy to advanced digital laboratory of James Johnston. In an apparent violation of the Nyquist criterion, this extends to non-repetitive signals like vocal consonants, finger snaps, hand claps, plectrum and percussion instruments. This is one of the main paradoxes where humans hear "mathematical impossibilities".

          The hearing processing also correlates echoes to the direct sound with this precision. Analog with infinite time resolution preserves the timing of the echoes, digital quantizes them effectively adding timing and level jitter every 22.6 microseconds. DSD, with 64X the time resolution of Redbook, is sufficient to reproduce an acoustic reverberation field.

          When you hear Belafonte moving, it is from the spatial characteristics of human voice, which are an approximate 90 degree cone from 300Hz to 3KHz, narrowing above and widening below. The timbre of the direct sound and echo pattern from the walls of the Carnegie stage tell your ears where he is, which way he is facing and how fast he is moving, exactly the information that is essential in hunting/gathering survival situations.

          This is how I can easily tell the difference between MP3, CD and SACD - the reality of the captured echo. Since it is completely missing from studio recordings, replaced by a variety of artificial means, I stopped listening to anything made in a studio.

          One exception: Master Performers Records, a boutique Australian label specializing in solo piano works, utilizes an electro-acoustic reverb by Steve Haas of SH Acoustics. This is an array of a dozen microphones and dozens of speakers covering all the walls that is programmed to emulate concert halls of different sizes and shapes using acoustic propagation, acoustic splitting and acoustic mixing.

        1. That's cool! That strikes me on paper as being a really good bang-for-the-buck sub system. 10" drivers only get you so far, though they can excite a lot of info in the average room.

          Glad he/they subscribe to asymmetrical placement. I tried some symmetrical two-sub placements in my room (fed by SEPARATE L/R feeds, natch), and it reminded me of wearing noise-cancelling headphones ๐Ÿ˜€ (read: flatter measured room response but awful/unnatural sounding, with wierd pressure on the eardrums)

    2. I heard Todd Welti and Sean Olive present the Harman International Audio Engineering Society paper on multiple subs, here is a consumer friendly version from the Harman website:


      Since 1978 I have been using a more advanced technique: bass leakage. Conventional architecture and interior design are not conducive to bass absorbtion nor diffusion in listening rooms. They reflect the bass waves to sum and difference in uneven geometrical patterns. The more symmetrical the room, the more asymmetrical the bass response - spheres, cylinders and cubes are the worst.

      The architectural antidote is letting the bass escape instead of reflect. Open doors and windows, stairwells, fireplaces, thin, light walls and sprung floors and ceilings leak bass into the environment and flatten the room.

      The next technique is eliminating right angles. Note that 45 degrees is not much better than 90, if at all. You need angles between 7 and 38, or between 52 and 83 degrees. Convex curves are also good.

      The best room geometry is a Gaussian diffuser at bass scale. These are also called Schroeder diffusers after the discoverer, Dr. Manfred Schroeder. Commercial versions are made for midrange and treble, they need to have dimensions of 5 feet depth to affect 50Hz.

          1. I guess you have to try both symmetrical and non-symmetrical placement and see if it works for you in your room in practice rather than on paper.

            Symmetrical results in heavy cancellations (like noise cancelling headphones, as noted below), which measures well to a microphone, and cancels a lot of the buildup around the room, but sounds unnatural and odd as hell to me.

            1. What worked for me is using multiple subs with adjustable phase. It tool several days to dial in as one adjustment changes all the others. You also have to adjust the low pass roll off of each sub. After a while you get the hang of it though. This has proven infinitely superior to all the other large one box solutions that I've had over the years. All I hear are bass instruments playing. They have texture and the sound of speed for lack of a better word that I've not herd very often outside a concert hall. I feel lucky to have been directed to this setup by others before me. FYI my subs are not symmetrical.

      1. Toole spoke of this in his book. He thinks the typical audiophile approach of two sheets of sheet rock glued and screwed is a recipe for disaster. I think he spoke of fixing a room by ripping all this out.

        I had the good fortune of building my house and did just about all of the things you are speaking of and I'm pleased with the results. Unfortunately I'm a rarity. The vast majority of us just have to make do.

        1. Exactly, heavy walls make frequency response worse. The funny part is Western European court music evolved in stone and plaster walled chambers. They did tend to have a lot of furniture with complex shapes, ceiling coffers and heavy fabrics everywhere, like floor to ceiling drapes, over-stuffed chairs and tapestries covering the walls.

          1. My room is pretty large. 18' by 27' by 11

            The side walls are book/ record shelves with only a 1/4 inch backing that opens into a space that goes to the outer walls which are foam insulated.

            The ceiling is a bit of a barn shape and all corners have been made to either not be 90 degrees or of so I have a short step to stop sound from bunching up there.

            Three of my corners have doors opening in to other rooms.

            The plan was to let as much bass exit the room.

      2. Interesting to learn about the architectural antidote, here. That's just what my room has and the bass sounds pretty decent down to the low 30Hz range, though it could use some augmentation from a couple (or swarm) of subs. Thanks!

      3. Wow this is odd. Lyric is one place that had wave ceilings and brick with wood as well as wood filled in between with fabric.
        My room was kind of designed around an audio manufacture
        It's a rectangle room all 90 degrees.
        I am not saying I know cause I don't.
        I read plenty and then asked him for advice. I did the best i could with his instructions.
        The building is brick and block the floors started as Cement slab.
        It's size is 20/9 by 55 deep.
        This is finshed dimensions
        THe walls are 2/6 wood insulted away from brick and block meaning they free stand
        Over that is 5/8 sheet rock over that glued to walls is 5/8 flake board
        The ceiling is the same hung from beams but has genie clips hundreds of them attached to half track then Sheetrocked and 5/8 Flackboard
        The floors are 5/8 soundboard called homosote it's used at ice skating rinks that have wood floors over them for events
        This has 3/4 100 percent filled plywood and 3/4 hardwood 4 inch wide oak. For now the walls are bare not sure what I want yet.
        I did read what you posted of un even walls and floors. But I was made to understand with the kind of speakers I have and plan to buy one day the rooms is correct.

        1. How big are the windows? The doors and corridors?

          5/8 Gypsum + 5/8" chipboard is lightweight wall, but with the studs are all the same spacing there will be a panel resonance. One of my best rooms was a finished basement. The previous owner had covered the rubble walls with 1/8" Luan panels, but the studs were irregularly spaced so each section absorbed a different bass frequency.

          Your floor is springy enough so it is less fatiguing to stand and walk on than the slab, but not springy enough help with bass.

          It sounds like you lost quite a bit of volume taking up 6" from the structural walls and dropping the ceiling. I would leave the ceiling open and use the cavities between the beams to make a bass diffuser.

          You mentioned Lyric - is this in Manhattan?

          You should come hear a concert at Spectrum, "the best place in New York to listen to piano":


          1. No Windows's on any walls besides two st the wxteme end
            There is a door that leads to outside it has chip board over it.
            The walls have beams at 12 on center he ceiling is 3/12 on 10 inch I did what I did for sound proofing I have TENET's a live my room. The room was almost 22 feet wide but to make it square I needed to loose a little. I lit bracing st irregular locations to do what toy mentioned
            When I sweep the room there is no resonance above 100 but some crazy stuff below. Moving them back in or out got me to get pretty good enough there is little to not boom anymore.
            I found out parts of my body vibrate at about 30 to 40 hertz. Also there is little to hear but it hurts my ears that low. If walk around I can feel and hear the bass change. Or why I sit back some 17 feet now. I never felt the floor vibrate but there was a bass bump. While I have no idea what right I do love it's sound now.

            1. 17' back the spatial characteristics of your speakers and echoes of your room will dominate and mask the acoustics in the recording. This is the trade-off in your approach. It may sound better than nearfield depending on your music and speakers, but it is very far from what it sounded like in the room with the musicians or the control room where the engineer and producer sit.

              You have flattened the mid-bass in one listening position with conventional architecture and tuning. The irregular stud spacing and compound wall (dissimilar materials) are the most important factors in your efforts so far. 5/8" sheetrock is transparent to frequencies below 100Hz, so that is where you lose control.

              The longitudinal modes are at multiples of 10Hz. To even this out, it is going to take interior walls or large, heavy furniture.

              You could also try the "Golden Mean" positions, which are 11'9" for the speaker face and 21' for the sweet spot, or the other way around.

              1. The just I got from this design as told to me is the 5/8 on the walls in my room. The wood is not screwed but glued . Where I sit is for best bass details but it does overall sound better at 15 feet .
                Now I am advised to sit a min of 15 feet away. To allow the sound to open up. With the right recti tube in my DAC I hear staging beyond the speakers. Also imaging that has height , width and great depth. My real,issue now besides the bass is getting the image to stay . It seems a male or femail voice moves at times not staying center. I also am aware of mixing changes . Drums , and other instuments seem to move at times. On reference music picked out for this purpose it's stable.
                Thnaks fro helping me. Maybe sometime we can meet for lunch in NYC .

  11. Awful suckout between 100 and 200 hz. Roy Alison researched and solved this problem over 40 years ago but even he knew more than this guy does. looks like it has a depressed treble too.

    Manufacturer's spec "19.5Hzโ€“33kHz, ยฑ3dB" Based on Atkinson's curve I'd rate it at 20hz-20Khz +4/-8 db

    Besides being an awful speaker, It is as ugly as sin. It looks like a matronly robot with a spinal deformative disease called kyphosis.

    1. Of course a 1/6 octave smoothing tells us little. Are the suck outs fairly wide which is easy to hear, or is the plot reacting to a extremely deep but narrow suck out which can be almost inaudible.

  12. Frequency response is the audiophiles mantra, and yet Gordon Holt's opposing dictum was that fr was not the most important. My research shows that audiophiles develop extreme hearing sensitivity to frequency response, but also develop a corresponding phase deafness. OTOH, musicians are insensitive to frequency response as documented by Harman Labs surveys, and very sensitive to phase. Atkinson doesn't measure phase of acoustic output (it would be bad for marketing), but does provide an indication of the degree of phase anomaly in the Alexandria XLF in this graph of step response:


    If you scroll down the page below to the samples of speech, you can clearly hear the effects of phase on the consonants. In the same way, speakers with anomalous phase response scramble musical consonants and spatial cues. If you feed a speaker a square wave, a microphone should show a square wave. Note that flat or linear phase on-axis is not sufficient criteria for reproducing music - speakers need controlled phase response at all angles.


    1. This is why I get bent when we are discussing fr-related things, yet a fr plot is taken as somehow being the last word. The current state of the art of our measurement is feeble and incomplete at best.

  13. Hey badbeef you may be on to something with that ear shape thing. Maybe someone can suggest the correct ear shape for acoustic Blues and I can start looking for a plastic surgeon. It will be cheaper than those Wilsons. ๐Ÿ˜€

    1. "Painstakingly reconstructed from the only existing photographs of Robert Johnson's ears, for the first time in History, you can now hear what the Blues Legend heard...."

      Let's start a company

      Combined with the Realizer A16 Headphone system, which does spatial cues by simulating part of your HRTF.....

      1. The problem with HRTF systems is that your response varies with small changes in angle. The only system I know that worked put a miniature pair of microphones in the ear canal, and locked the head in a vise for four hours while a speaker spiraled around in an anechoic chamber. That measurement and a supercomputer can do a convolution in real time to the necessary accuracy, assuming you have an anechoic recording with one track per instrument - and that screws up musicians' performance.

        I suggest concert tickets. Works for me. Best seats in Merkin Hall for five premiers last night, commissioned by one of the most precise chamber ensembles alive today, International Contemporary Ensemble.

        1. But isn't it a nice goal to be able to listen to the masters from the last century, plying music that you actually like, with the same amount of pleasure you get from a live event?

  14. Did you know about the frequency response graph of the Wilson watt puppys when you heard them ? Probably not yet they sounded great. In practice if a speaker sounds good the frequency response graph does not matter. All it does is make one aware of the irregularities and impressionable people begin hearing things they never heard before. On the other hand a speaker with close to perfect response on paper can sound not so impressive putting it mildly. The proof of the pudding is in the sound and that is why after making a perfect speaker according to the computer the final tweaking is done by ear called voicing. Regards.

  15. Here are a few unpleasant facts to consider.

    If you like phonograph records, your signal has been equalized many times even if the recording engineer didn't twiddle a single equalizer knob or push a slider. Master tape 40 db NAB equalization recording and playback. Mixdown tape, 40 db NAB equalization recording and playback. RIAA equalization 40 db recording and playback. Now if you have a recording made with Dolby A noise reduction the signal has been split into four bands and each is compressed to a degree based on its instantaneous loudness and the spectral distribution of tape hiss to maximize S/N ratio so the compression is non linear. Then it's put back together again and on to the tape. Then on playback the same thing happens only in reverse. That's eight more equalizers.

    Equalizers are what is known as zero phase circuits. When you correct the FR of a driver, you also correct its phase response. Multiway speaker systems are NOT zero phase systems. Each driver has its own phase response and group time delay. Combined they are a mess. Just look at that step function response again. The tweeter is done before the woofer even gets started. To make a multiway speaker system time and phase coherent, they would have to be coaxial and the signal to the lighter drivers with lower group delay would have to be retarded to coincide with the larger drivers in time with the longer group delay. Got any examples? I don't know of any.

    No matter what the on axis or combined power response of the speaker system is, the reflected sound will have a different FR when it arrives at your ears than the direct sound. This is because a) the speaker does not propagate uniformly directionally in space so the lower range of each driver has wider dispersion than sound at the higher end of its range and b) the reflective surfaces of your room don't reflect sound to the same degree at different frequencies. Any examples of speakers corrected for this problem? Mine.

    The famous acoustician Leo Beranek and Floyd Toole came to the same conclusion independently. That conclusion is that people like strong early lateral reflections. Anyone got a speaker that does that? No, the only one is the one audiophiles love to hate.

    Musical instruments propagate most of their sound energy away from any one listener, loudspeakers usually aim most of their energy directly at the listener. Now how can they possibly be even remotely similar? The obvious answer is they can't whether they cost $200 a pair or $200,000 a pair.

    BTW, the spectral balance of the source material is all over the map from one recording to another (and from one amplifier, phono cartridge, preamp to another.) Anyone got a system with a tool for correcting that? Oh you don't use those anymore? Too bad. If you know how to use them, they can make a good recording sound a lot better. But then audiophiles usually can't. In fact they are such powerful tools audiophiles do well to avoid them as they may get their fingers cut off by the blade. Stick to hand tools.

    1. RIAA and NAB equalization are conjugate phase, so the waveform is restored including phase. Speakers are more complicated, and do not.

      One reason drivers get out of step is inductance. Since sound pressure is proportional to cone acceleration, you need to jerk the cone hard to get proper transient response. Inductance limits jerk (third derivative of distance with respect to time). I use ultra-low inductance woofers and midrange (<300 microHenries) to get better coherence of step response. This is quite practical with Neo or Faraday.

      If the driver high frequency response exceeds Re/(2*PI*Le), it is from cone breakup and generally should be attenuated by the crossover.

  16. Regarding the Wilson Watt Puppies...another great speaker designer that's based in Colorado explained to me why the Watt Puppies got on my nerves. While I agree that they sound good...after a while...well basically he told me that the drivers were not totally coherent with one another. Yes, at the early stage of listening they do impress with their dynamics and punch but after an hour or more...About 7 years ago some rich guy who was selling a lot of his high end gear offered me a choice between a pair if gen 6 Watt Puppies or a pair of Andras by Eglleston works, either pair for the same price. I went for the Andras and never regretted that choice. In my opinion, speakers are the most ''personal'' in regards to choosing a system component. I rarely like to push any brand of speaker when someone asks me for advice of what speakers they should by. I find that personal taste play an important part compared to and amp or a source. So having said this, for my taste the Andras beat the Watt puppies. If anyone has had the chance to listen to a pair of Andras, they might have an inkling of what I'm getting at.

Leave a Reply

Stop by for a tour:
Mon-Fri, 8:30am-5pm MST

4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

Stop by for a tour:
4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram