The Amphion Krypton 3 is an unusual speaker with startling properties. If I were inclined to that sort of rhetoric, I would be inclined to call the listening results magical. Perfect it is not—it has some oddities of tonal balance, for example, and certain irregularities in frequency response of a localized sort. But it is all but unique in sound. HP used to describe such products as “demanding to be heard by any student of the audio arts.” That about covers it: If you are seriously interested in loudspeakers, you need to hear this one.
The reason is going to take a little explaining, so please bear with me through what may seem a digression.
There is a natural tendency to think about speakers in terms of spectral balance: Lots of bass or little bass, really warm or on the cooler side, midrange forward or midrange recessed, in your face presence or backed-off sound, bright or dark, airy and open or closed down in the top—there are lots of words, but they all come down to just frequency response of one band of frequencies versus another. Of course, frequency response does count in a big way, as anyone who has spent any time with a DSP or analog equalization device has found out in a hurry. It is the main determiner of the basic sound of a speaker. But the flip side of such overall frequency response observations with an EQ is that, while they are in fact crucial, they are not hard to alter. Some of the most important parts of speaker sound are controllable electronically.
But life is not so simple as all that, as just a one-dimensional thing, a “response curve,” whether you want to manipulate it yourself or not. How a speaker sounds in an anechoic chamber really is largely determined by that one curve, but in a real listening room, in the kind of room that you’re bound to have no matter how much work you have done acoustically, the sound of a speaker involves not just the on-axis response but also how the sound radiates into the room, what the response is on all the other axes, off to the side, up or down, in front and behind. It all counts. And it counts a lot.
All you have to do to hear how much it counts is DSP (or analog) EQ two speakers with really different radiation patterns to match on axis. They start to sound more similar than before most likely, but they will most definitely not sound the same. Make, say, a Harbeth M40 and a Quad 63, a box and a dipole, match exactly on axis—not that you would have to change either one all that much on axis—and they will still not sound alike. Too bad—because if on-axis response were the whole show, one could make a speaker do anything one wanted using DSP.
This importance of radiation pattern means that before designers start to work on the nuts and bolts of things, on getting the speaker flat on-axis and integrated and all the other things that need to be done, the designer has to decide what kind of radiation pattern to aim at. Possibilities are generally limited. Box speakers are omni in the lower frequencies, with a transition, the “baffle step,” to mostly forward at some frequency, usually somewhere between 300Hz and 600Hz, and they are almost entirely frontal in the top end, often with increasing directionality as frequency rises (“beaming”). Dipoles typically have a dipole pattern (no surprise!) until the real top end where typically they, too, narrow, becoming considerably narrower than the classic dipole pattern, though their backwave is still the same as their front wave but in opposite polarity.
But with enough ingenuity, new possibilities can be developed. Use of multiple drivers in the same range can increase vertical directivity, this reaching its apogee in the line source, in effect. And then there are some ideas that are really different. Conspicuous among these is the “flow-resistance enclosure,” introduced (as far as I know) by Jorma Salmi of Gradient in the Revolution model, back in the mid 1990s.
In this type of enclosure, the sides have many small holes cut or drilled in them that emit sound with a time delay so that it combines with the frontal radiation of the driver as that frontal radiation bends around to the sides. (Except at very high frequencies, sound bends around corners of enclosures—and of other things.) The combined effect is a cancellation of the radiation to the sides and to the rear as well. This type of speaker really does radiate almost exclusively forward and with a controlled forward pattern, too. You can check this by listening and you can check it by measurement: It really works.
This is exactly what the Krypton 3 does. Above the bass, it has a controlled radiation pattern from the flow-resistance midrange combined with a waveguide to control the tweeter pattern. And its radiation is controlled vertically by use of an MTM (midrange/tweeter/midrange) array. The control is enhanced further by the use of rather large midrange drivers. At the same time, the radiation pattern, while rather narrower than usual, is very nearly uniform over a considerable window in the horizontal direction. The off-axis roll-off is extremely uniform, above the bass with none of the down-sloping off-axis curves that have become common in narrow-front floorstanders—the off-axis is just as flat as the on-axis, just down in level.