Good afternoon everyone and what can I say after that? I'm going to try to be a little hasty here because times are passing and we have, as I say, some sound production coming after us here. Art and opinions are shown here in multiple colors and shades because that's really what they are.

It's whatever you want it to be. Art is whatever you want it to be. opinions are, whatever they happen to be. But I come from the science side of things.

It's black and white. It's reality. It's there or it's not there.

If it's good, if it's accurate, if it's repeatable, you can do it today, tomorrow, here or in Timbuktu and get the same answers. Science is different from art. Science is different from opinions.

And yet they go together. They must go together, in fact. If they don't, we will fail at our... our enterprise of entertaining the world with wonderful art.

Sound production, real thing happening on stage like this, it is art in real time and two performances may never be exactly the same. That's just what that's what it is. Sound reproduction, you put two letters in front of it, R-E, and the world changes because you're going to try to capture a performance, artistically craft it and save it.

so that somebody somewhere sometime can press a play button and if everything works right, they will hear what you've created. Sadly, that is rarely the case. Most of the time, what people hear in their homes, in their cars, through their earbuds or whatever, is not what the artists here or anywhere else created.

I think it's a shame. I think it's nice when you go to an art gallery, they make sure that the lighting is correct so that the colors on the paintings will be as close as possible to what the artist created. They don't put amber lights on or blue lights on.

They put neutral north light on. What's wrong with this picture? Well, it just simply says that a lot of research has been done on concert halls, some on workspaces because human health is involved, homes and cars and headphones. Not so much. But where do we listen?

It's the other way around. That's where we are entertained. We go to concerts, but not that often. We listen at home, we listen in our cars, we listen through headphones. I don't know how much, but certainly the trend is right.

And in fact, if you think about marketing of products and sound, I mean, if you think of... Look at that, I love that bit, because here in 1918 is Thomas Edison claiming that... he has created a reproduction device that will imitate a human being, or a violin. And this is a blind test. But you'll notice that some of the blindfolds cover the ears.

So it's a blind and partially deaf test, and maybe that's how he got away with it. But no, back then, and between then and now, there have been several... demonstrations of live performers beside loudspeakers reproducing live performers and the audiences would be amazed that they couldn't tell the difference.

Well you wonder you know have we made no progress in a hundred years since Thomas Edison? Of course we have. It's just that it's slick marketing, it's a good party trick let's just say that. In the case of the concert hall demonstrations, I might just add, I suspect there's good, strong evidence to indicate that it's the hall that makes the difference.

If they were to do that test in anything other than a really good hall, it wouldn't perform, it wouldn't resolve, it wouldn't work as well. Okay, if you get science involved, and we have to pay serious attention, we're going to do listening tests, they're not just sitting around with your drink of choice, with listening to your music of choice, it's actually hard work. It requires preparation and concentration. If you do it right, Under the right circumstances, you get something that I call subjective measurements because the results from the listeners are highly repeatable, very consistent with standard deviations of the same kind as you get out of technical measurements. If you're going to do technical measurements, they have to be accurate.

Otherwise, they lie to you and they have to be comprehensive. Otherwise, they're not telling you all of the truth. And so that is a requirement that...

eliminates an awful lot of people in our business from the game because they can't, they don't have the facilities to do these these highly precise technical measurements. I have been fortunate both at the National Research Council and at Harman to have had access to wonderful large accurate anechoic chambers to make measurements and dedicated listening rooms to do subjective evaluations. If you do both of those things and start comparing the results You very, very quickly see patterns.

They're not difficult. They're obvious. They just stand up and shout at you. That there are certain things you do in a loudspeaker, and if you do those, people give it the thumbs up right away.

They love it. If you don't do those things, then you're just taking the big gamble you don't want to take. So let's talk about listening tests first.

It's the single most important and really the most difficult factor in achieving this understanding of. the relationship between what we measure and what we hear. There are lots of opinions around, my goodness, just pick up any audio magazine, go on to any audio website, enter any forum on the topic, you will find opinions of all colors and flavors.

And they really just are noise in the background, and some of them may be right but you don't know which ones. So we need methods that turn opinions into facts. First thing we do is we make it blind.

It's obvious, you just disguise what the person is listening to. So they can't see the brand, the price, the size, and remember what their favorite reviewer said about it, what have you, that kind of thing. We make the test double blind. If you're a scientist, so then you know that you, the experimenter, or me, the experimenter, I am not influencing the result by the sequence in which I'm presenting these sounds, or anything is...

obvious as that. It has to be randomized and in fact the tests that we've done for years are randomized by a computer. We just press go and go have a cup of coffee. Test is done by the computer. So it is truly double-blind.

You have to make sure the test, sorry, the sounds you're comparing are equally loud. We do a multiple pair, multiple comparison test, so you're not just comparing one with another. It's certainly not the take-home and listening, listen to it kind of test that reviewers do. Because that teaches nothing. You learn really nothing from that.

Except that you might be able to adapt to its problems or not adapt to its problems. Only when you compare one thing with another do you start to realize what differences there may be. And if you do a simple A versus B comparison, and both A and B share the same problem, you won't hear that problem. So it's nice to have C and maybe D.

to mix things up a little bit. It seems obvious looking back on it. All I can say is that I'm very fortunate that the first listening test I ever did in 1966 or 67 was a three or four-way test and I've just been doing it ever since. But now we know that it's probably the preferred way of doing things.

So the listener controls which one of these speakers is active. when you're doing these comparisons, you do them one at a time, in monophonic. One speaker, not stereo, not multi-channel. Why?

Well, because it just turns out that monophonic comparisons are much more revealing. People are much, much fussier about what they can hear and like and don't like when they're listening just to a single loudspeaker than they are when they're listening to stereo. or multiple channels, multiple channels.

In fact, the more channels that are active, the less critical the listener is of the loudspeaker itself. But we have to pay attention to mono because it's a reality in television and movies. Most of the sound comes monophonically from the center front loudspeaker.

It is the most important loudspeaker in the whole room because all the dialogue comes from it, most of the on-screen sounds come from it, only in high-budget blockbuster movies do they bother to pan anything. So we do the tests in mono and we occasionally have done comparisons in stereo as well just to prove the point to questioning persons and every time. The ones that win the mono tests win the stereo tests.

There has been no exception to that rule. There is nothing special about stereophony or multi-channel So far as sound quality is concerned So here's a person one person at a time They just sit there and listen to music and press which of A B C or D they want to listen to They make notes the notes are compiled in the computers The music is in loops, so it's easy to make these comparisons over a period of time. Every time the music changes, the speakers associated with each of the ABCD codes also changes.

So each time the music changes, it's a whole new experiment. The person at no time knows what's behind the screen. Here I've removed the screen in the bottom picture, and you can see that there are three loudspeakers in that test. Only one of them is in the forward active position. It takes about three seconds to exchange.

It's all hydraulically driven and computer controlled. Cost a fortune. Howard Harmon paid willingly. Why bother?

Why do we bother spending that money going through all this rigmarole of blind and double blind and all that sort of thing? Well, because it matters. People are fickle. Humans are desperately fickle. We offer up opinions of what we're listening to, which are different if we see what we're listening to.

And to prove that point, when I joined Harman, nobody there could ever remember having done a blind test, not even a single blind test even. They were all professionals, I was told, you know, and as a professional, they're not biased by what they see. They really know what good sound is, and so... because they know what good sound is, they can do these tests without any complications of blind screening and so on.

So Sean had just joined the company at the time, Sean Olive, my colleague from NRC and McGill, now Dr. Sean Olive. Anyway, what we did was just to bring in some of these professional listeners and some others. and put them through a battery of tests.

Not everybody wanted to do it. There were a couple of holdouts, they just felt that they had nothing to lose if their reputation, you know, their reputations as golden ears were so high that they didn't want to threaten that reputation. Anyway, in these blind tests we had four loudspeakers. Two were rated very highly.

These were visually identical speakers. They differed only in their crossover designs. This, where's the arrow, there it is.

This test, the speaker down here, is a little $700 subset system. Disgustingly good for $700. Just disgustingly good.

It wouldn't play as loud as the other ones, but if you kept the sound levels at good foreground, listening level and not pushing it too hard, it just, it cooked. It was a good system for 700 bucks. These others are 5,000.

So here's, and the US competitor. Beautiful wooden cabinet, sculptured looking, wonderful, $4,000. And in the blind tests, we had two winners and two not so good ones. We let them see what they're listening to, and the loyal Harman employees gave these ones even higher scores, of course, because they were beautiful and they were big, and they were Harman products. This little one was a Harman product too, but they could never overcome the bias.

that it was so small and plastic that something small and plastic could not sound good and so they reduced its score. The competitor did a little better because they could see it now and it had this beautiful cabinet and so they gave it a little more but even then they felt that it wasn't quite up to the Harmon product. So you know we get different opinions when you see the product when you don't see the product. Another fact is that the room matters, especially at low frequencies, it matters a huge amount. Where you put the loudspeaker in the room and where you sit in the room changes what you hear.

And so if you move the speakers around and do the test, we should get different opinions when the speakers are in different places. Here on the left you see the black speaker. In the blind tests in two locations you get two very different ratings.

Huge difference. The white speaker, same speaker, two different locations, different ratings. This one, same speaker, different locations, different ratings. Same speaker, different locations, different ratings.

You let them see the speakers and suddenly those differences disappear. They say, well, it's the same speaker. It has to sound the same. I gave it a 5.6 the first time, I'll give it a 5.6 the second time.

They just kind of go deaf. Different recordings matter. I mean, recordings are not spectrally balanced in the same way, they don't have different, the same instrumental content, and so they will reveal problems in loudspeakers in different ways, and that's why we use a bunch of different recordings.

And here we have in the blind tests the black lines showing that the choice of recording affected the average preference rating, but in the sighted test, didn't make a difference. Play me anything. I'll give you the same number.

It's interesting, eh? And experience makes listeners fussier. And this has gone on for years. We noticed this a long time ago, that the people who are involved with the program and do a lot of critical listening, they're not as easily impressed as people without that kind of background experience.

And so the scores from the inexperienced listeners go up. And you'll see that... The males and the females agree. So anybody who entertains notions of sexism, you can forget them.

And this is not the first time I've done this kind of test. The listeners of both sexes are just fine, except that more males suffer from hearing loss than females, and so there's a probability that you'll have some faulty opinions from males with hearing loss. We just enjoyed louder hobbies, I guess. You know, so when you do sighted evaluations, when you know what you're listening to, even if you can't lay eyes on it, if you just know what you're listening to, your bias, your opinions are biased even before the first note of music has been played. If you believe something, you very likely will hear it.

And this happens, you know, those, anybody here is an audiophile or reads the audiophile press will know immediately what I'm saying. I mean, that's why these are the imagined, not real, imagined differences between speaker wires and that kind of thing. They're not real.

It's complete marketing. Is the room a fundamental problem? Well, one of the last things that was done before I left was to prepare this experiment and it was done after I left.

We compared three really good loudspeakers. They were closely rated. They were very good.

The differences were subtle. You had to pay attention to make the test a difficult test. And so we compared these three loudspeakers in three different rooms, completely blind. There was a screen.

The listeners couldn't see what they were hearing. And we randomized the positions and did it several times. It was all done properly balanced from a scientific point of view. And the upshot of it was that... In each of these three very different rooms, the listeners were able to rate the loudspeakers, and they rated them in almost precisely the same way, certainly the same order.

The numbers changed a little bit, but only a little bit. It's remarkable how small the differences were when you think of how large the acoustical sound fields differ from one another, how greatly they differ from one another. So we conclude that we have some ability to listen through rooms.

It happens in live performances. We don't have to equalize a Steinway or a Bosendorfer piano for which of the room it's in. It's the loudspeaker or it's the piano or whatever it is in the room.

And we can separate. the sound of the piano or the voice or whatever it is from the room to a great extent and When you think about what's going on that what has to be going on in this brain between our two ears It's it's a remarkable thing because Based on Masses at 40 years of doing this I can say that in order to give me or us anybody experimenter consistent, reliable opinions about the sounds that these people are hearing, they will have had to do these things. They will have had to extract from the sound field the timbral contribution from the loudspeaker so they can comment on the loudspeaker.

And to do that they have to separate the loudspeaker's timbre, timbral signature, from the room. And in order to make sense of either of those they will have had to take into account the timbral signature of the program, the recording. And believe me, nobody has ever, could ever claim to have been at the final mix of the recording they're listening to.

So how on earth is this possible? How can people, even inexperienced people, do this? How can experienced people do this? This is one heck of a party trick, really and truly. And in the end, after pondering this for years, it has to be, you know, Bregman, another McGill alumni, his auditory scene analysis.

It has to be that... we are streaming in our minds, we are separately streaming the sound of the musical instrument or the loudspeaker reproducing that musical instrument from the room. They're separate. If you go to a concert and the bass is weak, you don't blame the bass instrument or the musician playing the bass instrument, you blame the concert hall. So we are able to do that and we do it just subconsciously.

If my voice has a sudden timbral richness, you may recognize that we're in the men's room. And it's because of the reverberation in the room. It's still me on a scale of 10. The fidelity rating for my voice is going to be 10. But it's just Floyd in this room as opposed to Floyd in a different room. So we seem to do this, and this is an amazing feat.

Okay, we can do this. We can actually... listen to loudspeakers in rooms and offer up opinions on those loudspeakers that are repeatable from person to person from time to time and for different kinds of program material so we have a growing database of subjective measurements now we're going to create a parallel database of technical measurements so that we can start to stand back and compare one to the other hopefully figure out what it is in the technical performance of loudspeakers that makes some appealing to people at large and others that are not.

We have to admit right away that there is no single curve, no frequency response curve or whatever curve you choose to identify it as. That can completely describe the performance of a loudspeaker. it radiates sound in all directions into a somewhat reflective space, a room. And the sound field, by the time it gets to the listener, is extremely complicated. There's no way to unravel that complex structure.

You can't just put a microphone at the listening location and somehow analyze what it picks up and go back to figure out from that. what the loudspeaker sent out. It's impossible. You can't go backwards.

You can go forwards, though. You can measure the sound that leaves the loudspeaker in all manner of different directions and surmise from those measurements what would arrive at a listener's location in a room. So here's what we do. We do the spinorama that... Veislav mentioned, it's 70 measurements at 2 meters in an anechoic chamber.

We spin the speaker horizontally and then lay it on its side and spin it, get a vertical orbit. 70 measurements to measure the frequency response. In those 70 measurements we look at the direct sound, the on axis, it's the first sound to arrive at the listener in the sweet spot.

But we tend to want these days to entertain more than one person so we... do a measurement called the listening window, which is plus or minus 30 degrees, plus or minus 10 degrees, thus accommodating a reasonable audience, frankly. It works at home. It works even here, really. We measure early reflected sounds because sounds off the floor, the sidewalls and the ceiling, are the strongest, the second loudest sounds you hear in a room, are the first reflections.

So it's probably wise to pay attention to those. And to do that, what we do is we know the angle. communicates that wall bounce to the listening area, that wall bounce to the listening area, the ceiling and so on. So you measure those off axis angles and put those into the equation.

And to get the late reflections everything else you just do the sound power which is the sound radiated in all directions indiscriminately, just add it all up. And the computer will chew away for a nanosecond at that data set. and spit out a family of curves like this. The on-axis is the solid top line. Then there's the listening window, and ideally you want the listening window to be as nearly identical as possible to the on-axis so that the person in the sweet spot and the people adjacent get the same kind of direct sound because the first sound to arrive carries with it the essential timbral signature of the voice of the musical instrument.

And then we were interested to see if the follow-up sounds match that timbral signature, the reflections from the rest of the room, whether those are communicating similar timbral cues to our ears. So we look at the early reflected sounds, and this is the energy sum of the four seating side wall bounces and back and front wall bounces. Then we add up in a weighted sense, it's a complicated calculation, but we actually compute the estimated sound power radiated from the loudspeaker without regard to direction. We then compute two more numbers, the directivity indexes. DI, directivity index of zero, would be an omnidirectional source, sound radiated equally in all directions, and that is the case with all woofers and subwoofers.

Once you're below a frequency, once the sound source is small compared to the wavelength that radiates, it radiates equally in all directions. And so the wavelength at 20 hertz is 57 feet, and at 100 hertz is... 11.3 feet and the woofer is only a foot so no even at 100 Hertz the woofers are Essentially omnidirectional. They don't care which way you point them And so we calculate these directivity indexes as a function of frequency and they are really telling us something about The uniformity with which the loudspeaker radiates its energy into the room at low frequencies It's omnidirectional and as you go up in frequency it becomes more and more progressively directional facing the audience and then at high frequencies it will be quite narrow.

What we want is a loudspeaker that maintains something resembling a constant directivity over a wide range of frequencies. In that way and only in that way will we achieve this ideal performance of the direct sound matching the timbre of the reflected sounds. This spinorama, by the way, is now embodied in a standard published in 2013. And we'll see if anybody uses it. There are two problems with it. It's difficult to measure, not every manufacturer has that capability.

And even if they go outside of their own premises and hire a third party to do the measurements, which they could do, they might not like what they see. And so if they don't like what they see, you're not going to see it. That's really the fact of it.

So it's a challenge. But the data, if you can find best data, it's useful. Here's an example from my old original work, done in 1985, published in 86. Here's a loudspeaker that had some directivity problems.

It was designed, I know the designers, and I know the history of it. It was designed by a group of engineers who were smart, who were told to get the flattest, smoothest on-axis response they could possibly get. And that was the only instruction they were given.

And they did it. They went out, I'm sure, to the local pub and had a few beers to celebrate when they did it. But what they ignored, because they weren't told to pay any attention to it, was what happens off-axis.

We get this pristine first sound arriving at the sweet spot, but 60, 30, 60 degrees off axis? No, not so, not so good. So what does this mean in terms of how it sounds? Well, in terms of how it sounds, I can tell you that when we put this one into a double-blind listening test, it didn't win and it didn't lose. It was a good, solid, mediocre loudspeaker, which was, it was an expensive high-end loudspeaker.

It was a KEF 105.2 for Those of you who might have gray hairs and can remember Which was highly regarded at the time So here is the direct sound that arrives from that loudspeaker At a listener, that's the energy sum of the first four reflections. Only the first four reflections. Floor, ceiling, sidewalks.

And already we can see in the measurements why that loudspeaker had a personality. Why it wasn't neutral. Why it just sounded not quite right.

We didn't give it our highest rating. We didn't give our lowest rating because there are worse loudspeakers than that. But it's not perfect by any means. And when you go to the sound power, the total sound radiated in all directions, you get that.

And you can see that in the total sound power and in the early reflections, this inferior off-axis performance is the dominating factor. The fact that it's beautifully smooth and flat on-axis is not a factor at all. In fact, you think about this, you look at that set of data, and you ask the question, if we are to look at a data set and try to predict how a loudspeaker will sound in a room, what do we have to measure? Well, at low frequencies, we're going to measure sound power.

That's 10 dB higher than anything else. At highest frequencies, we can rely on the on-axis because everything else has been attenuated. The on-axis is the dominant factor at high frequencies.

because the tweeters are directional and the room is absorptive and air is absorptive. But once you get below about 10 kilohertz, everything gets into the act. You can ignore absolutely nothing. And the on-axis response is not the dominant factor.

So let's add those up. And we get a room curve. Now, what we see in that predicted room curve is evidence that, as a function of time, something is happening.

The direct sound, the first sound to arrive at your ears will be this black curve right here, the black one. But as a function of time, as those reflected sounds arrive at your ears, they're going to add up to that red curve at the top, the prediction. So now we see that what we measure in a room is dominated by the off-axis performance of the loudspeaker.

And the room itself, because the room has to communicate that energy to us, and if it doesn't communicate that energy equally at all frequencies, it can't be neutral. So we have two issues here. That's an interesting test.

Let's just say, let's just see how that prediction works. So we put that loudspeaker into a listening room, put it down in a typical stereo left location, measure it. It's averaged over those six seats.

Averaged over the six seats. Move it, we only moved it half a meter, 18 inches. Not very far.

But look at the difference in the curves. Wow. You either had a kick drum that's a little fat or a kick drum that was almost absent. It's not subtle.

That's what standing waves in a room do at low frequencies. And if you play around with the positions, you can find another position that looks pretty good. And if you listen while you're doing all of this, you find, yeah, that's the one that actually sounds like a pretty tight, good kick drum. I like that.

Bass notes are all there. And there's our prediction. So what do we conclude?

Well, first of all, the prediction worked. Not at all frequencies, but above 300 or 400 hertz, I'd say we nailed it. So there's enough information in the anechoic data of that loudspeaker to be able to predict.

What's happening in a room above three or four hundred Hertz? What happens at lower frequencies? Well, those are the very familiar room resonances, the standing waves, the modes, the eigenmodes.

People have many names for them. They're the corruption of the base in all small rooms of any shape. There is no escape from them, so you have to deal with them.

And if I have time, I'll show you how. So the prediction of that room curve is that... tells us that the shape of the room curve, and I'd have to say, probably what we hear substantially is dictated by what happens off-axis. It's reflected energy.

The only way to get rid of that reflected energy is to listen outdoors or in a very anechoic room, neither of which is very pleasant for listening to music, right? Rooms are good for music, even good rooms for... about speaker reproduction in music.

And in this case if you do a room measurement and see that red curve up there and you reach for your trusty equalizer and smooth it out what you've done is you've screwed up the only good thing that speaker had which was its on axis response. You have not solved the problem. You can't solve that problem with an equalizer.

The only solution for that problem is a loudspeaker that behaves itself. So you take that one back to the shop and buy a better loudspeaker. Like that one.

That's a better loudspeaker. Smooth on axis and quite well behaved off axis. And that's not a bad. The directivity index is, you know, it's not perfect but it's not horrendous. You put it in a room and you measure 6 seat average, you get that.

Now there's the sound power and the inverse of the directivity index, if you turn the directivity index upside down you will get an estimate of the sound power if that loudspeaker is equalized to be perfectly flat or created to be perfectly flat on axis. Now this one was almost perfectly flat on axis to begin with. And so the inverse of the directivity index and the sound power that we actually measured sit one on top of the other and if we sit move them down They give us a pretty good picture of the room curve above 200 Hertz Only when you get up into the you know above seven or eight kilohertz does the tweeter Does the tweeter start to dominate and the sound power curve doesn't wait the tweeter tweeter output sufficiently so we have to We have to add some more information, but I can tell you that what we're seeing is that over most of the critical part of the frequency range, we can predict a room curve from the anechoic measurements. It's not a mystery.

There's another one. Now, this is not a perfect loudspeaker. This is a modern version of what I just showed you.

In fact, I heard a pair of those today. They're here. the on axis is not flat and smooth, the off axis certainly isn't flat and smooth, the directivity index is not smooth, but because the on axis is not smooth then we will see that these curves are not the same. And let's just see, the sound power now should be the best predictor of the room curve and by golly it is.

And so the inverse of the directivity index is not bad. If you had nothing else to work with, it would be pretty good. All I'm hoping I'm developing in your minds is the idea that these cold scientific numbers and graphs that we measure in our anechoic chambers are quite relevant to what we hear in rooms. They're not abstract. Now, that loudspeaker has two problems.

It's not flat on axis. And an equalizer can fix that because an equalizer just changes the amplitude response, the frequency response. But it has problematic directivity and an equalizer cannot fix that.

And both of those contribute to the fact that the room curve is not flat. If you don't know that top data about the loudspeaker and all you do is measure the room curve, which most people do, A lot of receivers will do this automatically for you. You press the button and it makes funny sounds and reassures you that it has calibrated your loudspeaker and room's combination.

Well, that's what it's measuring and it has no idea what's wrong with the loudspeaker or what's wrong with the room. Doesn't know what's causing it, so it doesn't know what to do to correct it. In this case, equalization is not going to solve the problem. Only a better loudspeaker will.

Now these are popular terms. I don't know, does anybody here have a room equalizer in their home system or business system? Yeah, had to be one, thank you. No, because if you go to your local, if you order it by any of the modern-day domestic surround sound receivers, many of them have these algorithms built into them.

You can go out and spend hundreds or a thousand or more dollars for a standalone box that does this for you. Or you can pay somebody to come in and do it for you. And it's a system. It's a service. And the claim is that by making such measurements on unknown loudspeakers in unknown rooms, that we can create perfection for anybody, anywhere.

And unfortunately, there are some standards from SMPTE. SMPTE. motion picture industry, ISO, ITU, that actually promote this belief. And I'm afraid to say it is simply not true.

It can't be true, and it isn't true. And in my dying years, I'm going to do my damnedest to change it. The fact is that when you consider all of the audible characteristics that two ears and the brain can extract from the sound field of sound source in a room, it's not going to be accurately represented by a microphone and a simple analyzer.

It's just a presumption that isn't true. What we're measuring is not what we're hearing. So back to now what we are measuring and what we are measuring is that the loudspeaker is in control over part of the frequency range, the room is in control over part of the frequency range and the transition frequency between the room and the loudspeaker control varies with the size of the room and as the room gets larger the loudspeaker takes over more and more control.

So in a room this size, the loudspeaker is going to be in control down to probably, oh, 40, 50 Hz. And below that, we're going to have to pay some attention to this particular space. But above that, we have some assurance that what you're going to hear will be dominantly influenced by the loudspeaker.

And so, in theory, if I'm right... All I need to do is to put a really good loudspeaker here, turn it on, and you'll be a happy camper. Except at low frequencies.

We may have to pay some attention to low frequencies. Now, the interesting thing is that I was thinking along these lines, and then I stumbled into a man named Art Benade, a physicist, musician, brilliant man. He's dead now, but here he died. He sent to me a collection of his writings.

And this was among them. I said, well, this is it. Because he just said, well, look, as a musician, it's obvious.

The musical source is well-defined. I mean, I can play the same riff ten times, and it'll be the same acoustical output ten times. And back in the audience, it sounds the same ten times.

But if you make measurements, and you walk around the audience, and it still sounds the same. But if you make measurements from here to there... It's going to be very different.

You put the microphone in different positions, you find huge, huge, huge variations. He's suggesting, or my interpretation of his suggestion, is that if we're able to describe the sound source in measurements, then we should be able to predict what listeners are hearing. What happens in between...

maybe doesn't matter quite so much because they're going to be very very difficult to interpret. I mean if you make measurements in a concert hall and put a sound source on the stage and measure back in the audience and do a frequency response transfer function what have you it's going to be awful awful awful awful. An engineer will look at it and say this is just a staggeringly complicated comb filter can't possibly sound good.

Communication may be impossible. and yet we go there and we listen to it and we love it. It's a reference listening experience. That's the difference between what we are measuring and what we are hearing, and in this case we are measuring the wrong thing.

So let's just draw the parallel between live performance and reproduction. We have a musical instrument plus room to the brain. The brain sorts it all out.

Now if we have a good recording of that instrument, reproduced by a good loudspeaker, we should be able to... sort of have a parallel experience in our homes. And that's the challenge. That's what we're going to see if it happens.

I'll tell you it does happen, actually. But the success of that whole exercise will depend on the competence of the loudspeaker designer and on the ability of the recording industry. That's you people, some of you anyway, to capture and store accurate renderings of what... was the sounds that were radiated by those voices and musical instruments that we care to listen to.

And this is an artistic exercise and you just go into any one of these control rooms and you see these massive consoles and off-side apparatus with knobs and dials and all kinds of processing available and that is perfectly legitimate. But you know, it's legitimate and it's part of the problem. the art shouldn't be whatever the artist wants it to be. It's just that we have trouble delivering that art to the customer. See, we listen in homes, in our cars, and through headphones.

That's the audio industry. The audio industry is you capture sound, you process it, you send it off to a live audience in real time. You can store it, you can send it off to customers wherever they want to press a play button.

Or you can send it off to a different form of storage and send it off to the... cinema world. So that's the audio industry as I know it, as I've worked in it all of my professional career. And my philosophy is embodied in this, science in the service of art is our business.

And when you combine those three words it's kind of oxymoronic. Science and art together? Yeah.

Business? Science and business? No.

It doesn't work. Art and business? Music, yes. But good sound is our product. I mean, I worked for a manufacturer that made boxes that cost money.

Wooden boxes, steel boxes, but it's the sound that's the product. That's what puts the smile on the customer's face or the frown on the customer's face. You know, if we screw up, the customer's not happy.

And it's not happy because the box doesn't look good. He's not happy because the sound isn't good. So how do we know what good is? We listen. And when we listen, we get caught into what I call the circle of confusion, where loudspeakers are evaluated using recordings.

And those recordings were created in control rooms where you play with microphones, select microphones, position microphones, play with equalizers, add spatial temporal effects and what have you, while you are monitoring what you're doing through loudspeakers. And so you know where this is going, right? I mean, the loudspeakers aren't standardized.

Over the years, they have varied horrendously. And I've recently even been into control rooms where I wouldn't give house room to the speakers that they are listening to. They're creating art through really dreadful products, dreadful loudspeakers. And as a result, the recordings are variable.

I mean, I was talking to this just not long ago saying, you know, I sat down with headphones on, which I don't normally do very much. I don't really care much for headphone listening. But in this case, I was going to explore streaming music sources. and see how they were doing for keeping the sound levels, the playback levels, relatively constant from tune to tune.

They do that quite well sometimes. But they also... I realized immediately that the timbral balance, the spectral balance, was varying hugely. So, I mean, I went into the algorithm, played with the bass control, played with the treble control, and found that really quite a high percentage of the recordings I was listening to benefited from tone control adjustments, those old-fashioned things, bass and treble, that my daddy's apparatus used to have. You know, it's because they are variable.

Well, why are they variable? Part of it at least has to be the monitoring circumstances, because if the monitor speakers are radiating too much bass, you don't put as much bass into the into the disc, into the recording. And on and on it goes. I mean, you can see where this is not very, it's not heavy medicine here. We have to have some relationship between the domain within which the art is created and the domain within which the art is appreciated.

These professional monitor loudspeakers in rooms have to behave in some ways similar to our consumer loudspeakers in rooms, or cars even, or the system doesn't work. The art that is being so carefully created by the performers and by the recording engineers That art is not being communicated to the audience in a way that allows the audience to say with any certainty that if I don't like it, I don't know who to blame. Do I blame my loudspeakers? Is it my room that's the problem? Or was it just a crappy recording?

And that's the dilemma you're in, unless you have measurements. And so the... We can anticipate the sound field in a room, the physical sound field in a room from that spinorama data.

The real challenge now is does that translate into what we perceive? And so one day I get a phone call from the CEO of Harman saying, Tool, we pay you a lot of money to make sure our products sound good. Yes. Well, can you explain the recent or the current Consumer Reports review of loudspeakers in which some of our products did not do well? Well, I said, can I talk?

So we went into his office and we talked and I explained to him that for decades, those of us in the loudspeaker business knew that Consumer Reports reviews didn't really mean anything because they were doing it wrong. Well, he said, can we teach them to do it right? And I said, yeah, probably. It's going to take some money and some time.

And so he authorized it. He just says, okay, do it. Make it so. So Sean Olive and I got together, and we put together what we knew about the psychoacoustic relationships between these curves and numbers and subjective opinions. And we went back to our database because at Harman we do these listening tests routinely and we do the measurements on our own products and on our competitors'products so we know where we stand.

And at that particular time we had a database of 70, on 70 loudspeakers. We had these double-blind single-listener. ...meticulous listening tests where the result was a score on a scale of 10. Now that's listening to a loudspeaker in a room, so that's the combination of the loudspeaker and the room that's being rated.

Part two of the test is to take those loudspeakers into the anechoic chamber, do the spinorama, and process that spinorama data through an algorithm that tries to pay attention to spectral balance on and off axis. bandwidth, narrow band, timbral problems, resonances, and what have you, and come up with a number. So we have a calculated number based on the anechoic data, and we have the number that came out of the listening test, and the correlation was 0.86.

That's not chabby. You go to Las Vegas with those odds, they'll kick you out of the casino in no time. But going back to the origin of that, the motivation for this, Consumer Reports, so they're always, there's the pretty perspective on it from their anechoic data, we've actually predicted listener reactions to the loudspeaker.

But we did 70 loudspeakers and in that 70 loudspeaker collection, there were big expensive speakers and there were little inexpensive speakers that had no base and had no pretense to excellence. perhaps, but could be quite good over a smaller bandwidth, narrow bandwidth. So we separated out from that 70, 13, the original 13 that Consumers Union had published, and there's the result.

On the left, these big dots here are the results of the listening tests, our double-blind listening tests, from the worst down here up to the best up there. The square, the open squares are the consumer reports ratings. And you can see our best is their worst.

And this is fascinating. It was a slightly negative correlation, but with a very low probability. So there's a lot of variation.

And when we put it through the algorithm calculating algorithm that Sean had created. the correlation was 0.995. Well, that's one.

That's perfect. The calculated subjective ratings and the real subjective ratings just agreed. You can see they ran on the same line.

Here's a very specific example, kind of an interesting one. There are four loudspeakers. These were all high-end, audiophile-approved, highly regarded, reviewed loudspeakers. If you follow these magazines and read these reviews, you would feel quite confident in going out and if you had something of the order of $10,000 to spend, which is what these cost, each pair, if you had that kind of money to spend, you would feel quite confident that if you bought any one of them, you'd be buying something resembling the state of the art.

But in our listening tests... They weren't all state-of-the-art. There were two that were in a close horse race for first place, and the other two were just further down.

And that's the spinorama data. And as they get flatter and smoother, the scores get higher. It's not magic. We would never, ever buy an electronic component. Nobody would ever design an electronic component that didn't have a frequency response that was flat.

10 Hertz to a hundred kilohertz of you know that kind of thing is flat and smooth It's not not difficult to do and everybody does it Why would we think that when you get to a loudspeaker we can kind of throw that rule away? Well it turns out that what's what our listeners here are doing is saying no you shouldn't throw that rule away It's a good rule do it And if you do it your products going to sound better. I will like it more Looking at the...

there were 350 listeners that participated in that test over three months or so period. 268 had done the test when Sean wrote his paper, it's in the Journal of the AES, showing how the listeners'performance of experience and background related to their performance. And their performance here was based on the FL statistic.

It's a measure of how consistent people are when they experience the same loudspeaker on different occasions because we do multiple times, they hear the speaker several times in different orders, different sequences. And on the strength of their differentiation, if they hear some loudspeakers that are from good to bad, and they give the good ones sixes and sevens, and the bad ones twos and threes, we know that's strong differentiation. But there are people who don't do that. They will give you sixes and sevens and fives and sixes.

They don't really spread it out. It turns out that the ones that exhibited both of these characteristics most are selected and trained listeners. First of all, they're selected on the basis of having normal hearing.

Not everybody has normal hearing. So that's the first hurdle they have to pass, and then they are actually trained to recognize different kinds of defects. The retail salespeople did quite well.

The audio reviewers, six of them who volunteered to come in, they were bold people. They didn't do great. Our own brand and sales and marketing people did really poorly. These are the people who are going out into the field, say, buy our product, it's great. I'll vote for it.

Please don't. And then the students, these were not students like you people in a music and sound recording environment. These were just generic students. Non-audio, non-music students. Anyway, it's kind of interesting.

The point is though, they differed in their ability to give consistent answers and they differed in their abilities to give numerically spread out answers, but in the end they all agreed on the ratings of those products. There were no inconsistencies. It was really remarkable. Humans are amazing measuring instruments. Amazing.

I mean, I've done this for 40 years and I still kind of say, wow, this is remarkable. Because as I say, you're listening to strange music in a strange room, to strange loudspeakers, and you're forming opinions. And the only way we can get those consistent opinions out of you is to not let you know what you're listening to.

Do it blind. But if you know what you're listening to, I don't care what you think. It doesn't matter. And so, you know, I go back to Art Bernays'idea.

He was absolutely right. Absolutely right. The difference is that we can measure loudspeakers that are reproducing musical instruments. We don't have any parallel way to measure the musical instruments themselves because they are so different and so varied. so peculiar in the way they radiate sound into rooms.

So we have those two measurements now and we combine them and that gives us some psychoacoustic relationships and what we learn from that is that to make a good loudspeaker it has to be smooth and flat on axis. If you don't, people will complain. The directivity index should be smooth and relatively constant.

It can change, but it should change gradually and smoothly, it seems. We're still in the learning process there, but it looks pretty good. We don't want any resonances. We've got to find in our measurements ways to identify and get rid of resonances, and we can do that in our spin-arounders. And the reason why resonances are so important is because they are the building blocks of every sound that we care to listen to.

Human voices and the sound of my voice is a bunch of resonances. Musical instruments, just bunches of resonances. And so if you find additional resonances in loudspeakers, you're coloring the sound.

And people recognize that quite quickly because that same coloration shows up in every recording, every music, every voice you play. So we want to get rid of that. So good loudspeakers are flat, probably constant in their radiation patterns that were the frequency range.

They don't have resonances. In order to do that we have to measure at high resolution. The old idea that third octave data is useful. I'm afraid if you have a third octave analyzer give it to your kids because it's a toy.

And bass performance. This is fun. This is a big deal.

Interesting. Bass performance all by its nonsense. accounts for about 30% of the factor weighting in the subjective evaluations.

And if you don't get the bass right, everything else kind of suffers. Getting the bass right is a bit of a trick. When you think of loudness, loudness in the middle frequencies changes by a factor of two for each 10 decibel increment. You know, middle frequencies from two or 300 hertz to...

3 or 4 kilohertz roughly. If you want, if you turn the volume up 10 dB, most people would say, yeah, it's about twice as loud. You turn it down, it's not half as loud.

That's a factor of 2, 10 dB, except at low frequencies because the low, the low, equal loudness contours, if you go back to your textbooks, equal loudness contours crowd together at low frequencies. And so at 50 Hertz, it may only require 4 or 5 dB to produce a factor of two differences in subjective loudness. And so if the bass is off by a little bit, it reflects badly on the rest of the whole spectrum. So getting it right means getting it right.

Can we predict any of this from manufacturer's specifications? Because it would be kind of nice if we could. I mean, it's fine for me to stand up here and tell you...

that if you have a Spinarama, you can find out all kinds of useful things, predict how it sounds, predict how it measures, very cool stuff. But you go shopping for loudspeakers, you don't see Spinaramas. That's going to change slowly, but right now that's the case.

You find a few. There are some of the Harman products that have Spinaramas. You can go on the website and find them.

The M2s that you guys just bought have their Spinaramas up on the website. Anyway, for $200 a pair. You don't expect a lot.

You go to your local box store on a weekend and they have it on sale for $150 a pair. And you put a pair under your arm and go home. And all they're telling you is that the performance of that loudspeaker from some 58 Hz here to 22 kHz up here, it's going to fall somewhere in that tolerance range, plus or minus 3 dB.

6 dB tolerance range. 6 dB, gee. How many dB to double the loudness?

Did I say 10? Well, 6 is not that far away, is it? Now, if you see a curve, it shows you a lot more information. You get a lot more confidence looking at that curve. It's tilted, and you can bring that down with the tone control tilt.

You can flatten it out if you wanted to. It'd be easy to do. If you knew that curve, if you had that curve, you could think about doing that. But if you don't have that curve, you can't think about doing anything because the manufacturer has not told you anything useful. If you have the Spinorama, wow.

Whole different ballgame Now we can see that it has resonances How do I know those are resonances because they are present in all of the curves all the spatially average curves show the same little bumps and Those are resonances and loudspeaker transducers are minimum phase devices now that is really technical jargon But I'll quickly explain it means basically that if the frequency response or the amplitude response is smooth and flat or smooth no bumps, then there are no resonances. If you see a bump, there will be a resonance. It will ring.

And if you have an equalizer that you can match to those bumps and get rid of the bumps, then the resonance is gone. There's no more ringing. You've corrected the problem.

So this is one of the reasons why active loudspeakers with built-in amplifiers and digital electronics are, I would hope, the future. Because that's the only way we can really make super good loudspeakers. The M2 is one of those. It's dedicated electronics.

$1,800 a pair. Looks pretty good. Better than the spec, actually. But here's the case where the marketing department said, ah, plus or minus 3 dB industry standard.

Everybody understands it. And it's a giveaway statement. Because it does a disservice to this extremely good loudspeaker. So for $1,800 a pair, if you were fortunate enough to go out and find those and buy them, you ended up at home with something that is dangerously close to the best that can be done, period. Then, certainly.

Dangerously close. So if you were to ask me what the point of diminishing returns would be, clearly that's an example of it, because honest to God, you could pay ten times that, or more, and not get as much. in terms of timbral excellence and accuracy as that portrays. Here's the 11,000th pair.

Whoa, whoopsie-doosie. Okay, so that's not great. If we could cheat and smooth it out and it would fit in there maybe, but we don't like to do that. We show the whole spinorama, and what we see in the spinorama is that we have resonances, folks. Now this, just out of curiosity, this is a...

a large electrostatic panel sitting on a standard woofer. Looks cool. Very sexy, very industrial, good industrial design.

People just fall in love with it and fall all over themselves. And then the technical stories come out, well it's an electrostatic loudspeaker and therefore the diaphragm has no mass. That means it moves instantaneously, its speed, it's fast, it's accurate. Well no, it's not. It's not fast, it's not accurate and it rings like kick at a whole bunch of frequencies because It's very directional.

If you look at the difference between the black curve and the green curve, that's the on axis, that's the listening window. That means that no two people, even on a sofa, will hear the same sound. It is truly a bad design.

And it's a poor investment, obviously, at 11 grand a pair. Now you can go up to 24,000. Whoa, rock and roll. Okay. That's that.

It uses up the whole plus or minus 3 dB tolerance. You'd say, well, for 24 grand maybe they should have done a little more work than that. It meets the spec, but I wouldn't give it any prizes.

Look at the spinorama. Well, my goodness, what have we got here? We've got a resonance!

For 24,000 we have a resonance that we didn't see in an $1,800 loudspeaker. Isn't that interesting? You can hear that because it's above the threshold of audibility and it's not flat.

It's got a directivity problem a little kink in the directivity it radiates widely and then all of a sudden it gets decides to get narrow It's not flat anywhere. You can't equalize it So why bother? Why why would you spend that money for that loudspeaker?

16,000 to pair now here's this is an interesting. It's plus or minus 1 DB Now that's getting serious. That's an audacious Challenge and they just squeak it, just squeak it, barely, boy. That's a brave thing to do.

And I'll tell you why it's a brave thing to do, because one of the fundamental problems of the loudspeaker industry is that... I've got to quit...is that the consistency... It's one thing to design a golden prototype. and you might get lucky and do a really good job on that.

It's a very different thing to clone that in mass production. And when we do some products, this is one where we guarantee that it's going to be a win in spec, and the only way we can guarantee that is to spend a lot of money in the production process. That's part of what you're paying for, is consistency.

And it's sad because there are products that from other sources and other brands and what have you, that you don't know what you're buying because they're small companies, they don't have any real production quality control, and so on. It's an insidious thing. That does sound good, by the way.

That's reference quality sound. And just a joke. There's more useful and reliable information on the side of a tire than there is in a loudspeaker specification sheet. And this is not true only of consumer loudspeakers.

It is equally true of professional loudspeakers. I go on the websites and I check out what their spec sheets are telling you. It's nothing.

They're not telling you anything useful. They're saying, trust me, trust us. That's the M2, by the way.

It's arguably about as good as speakers get these days. And so how does it sound? Well, depends on what you play through it because of the circle of confusion, right? And if you have a perfect loudspeaker, that's a pretty good example of where we stand at the moment and if we get even better than that in the future and we have somebody somewhere creates the perfect loudspeaker, is it always going to sound good?

Hell no. Because the recordings are so variable. The recordings are all over the place.

We have no idea what kind of loudspeaker or room or what circumstances under which they are mixed. And so when we hear what we hear through unmeasured loudspeakers, we don't know where the problem came from. We might like or dislike.

We don't know who to praise. Do we praise the loudspeaker because it sounds good? Or if it sounds bad, do we condemn the loudspeaker?

Well, it might have been the recording. Only if we have neutral loudspeakers in the production environment and neutral loudspeakers in the home reproduction, car reproduction environments, will we get out of this unfortunate circle of confusion. We have to get people mixing on good speakers and people at home listening on good speakers. I'll have to tell you right now... People at home are doing pretty well.

Even cars are getting better. I don't know whether you've noticed. Even some of these little Bluetooth connected wireless loudspeakers don't sound half bad. Things are way better than they were 10 years ago, certainly 20 years ago. So good sound is coming, and it's coming partly because of efforts from people like myself who go out and find out how to make measurements and use them and then teach it.

and standardize it and hope people will pick up on the idea. But in the case, if you have good loudspeakers and you hear something you don't like, you know the problem is in the recording. It's not in the loudspeaker. That's the thing. That's the thing.

I used the example earlier with Vyoslav. We were talking in his office, and I think it's like that clear window, and you, the recording people, are creating the scene outside the window. And if you don't like the scene, the window's clear.

If you don't like the scene, change the scene. That's your job. That's you, the artist.

But if the window's pink, you have no way of knowing. And if I have a blue window, and what have you, it's a silly analogy. But no, it's not a silly analogy, because it's real.

It's the same thing. When you have a neutral loudspeaker, you may not even like recordings you've made yourselves when you listen to them. But that's fine. That's the way it is. You weren't listening, perhaps, through a neutral loudspeaker when you made the mix.

That's the thing. So from here on, though, you have the opportunity. You have the opportunity to use neutral loudspeakers.

And at the present time, they're getting down in price to the point where, as I showed, even 200 bucks plus a subwoofer gets you in the ballpark. Only then can we create art in the circumstance where... it might be reproduced and now fortunately we are able to measure neutrality.

I could end there. I think I probably should this time, right? Okay, thank you. Actually I had one more slide.

I have one more slide to show you. I forgot I put this one on. This is the final, this is the cap, this is the last slide.

Here's the M2. That last, I'm sorry? Now it's showing, okay. That's the M2 measured in a, the red curve was measured in one of our home theater rooms at Harmon. 6 seats, just a small, typical domestic home theatre.

And the other curves are measured in film sound facilities ranging from 24, 60 seats, 114, 516 seats. Pretty big. Measurements are the same, up above frequency, above where the loudspeaker takes over. Below where the room takes over, you've got a different problem.

You've got a different problem. that low frequencies, that's the room getting in the way. And we have ways of working on that.

I can talk about them tomorrow. But above, the good news is that above about 200 Hz, we're done. And I'll have to tell you that these little irregularities in the cinema, the black curves here, those ones that stray out, those were the ones that were measured with the microphones at ear level. which puts the microphone dangerously close to the back of the seat, which is an acoustical interference thing, that the human two ears and a brain, we don't hear it. You can stand up and sit down in a cinema sometime or in a concert sometime if you want to be impolite.

And you don't hear that kind of thing. But you measure it, and if you measure it, the temptation to the technician is that they will equalize it. And in equalizing it, they're equalizing something that cannot... should not be equalized and they're making what might have been a really good loudspeaker sound worse.

So now I am through. Thanks.