The French company Devialet has introduced a number of amplifier innovations since its founding in 2004. It is most famous for its hybrid output stage, which uses a Class A amplifier for voltage gain and a Class D amplifier for current gain. Devialet’s amplifiers look and operate differently than conventional products, combining sleek industrial design with very advanced software-driven features that both improve the sound and make the products easy to use.
The Expert 220 Pro integrated amplifier reviewed here incorporates many of the more than 100 patents the company has been granted over the past decade. One of these patented technologies is of particular interest to me: Speaker Active Matching or SAM.
Addressing a fundamental problem of loudspeaker bass behavior, the SAM system is a new way of correcting transducers in the lower frequencies. A truly exciting and spectacularly effective device, it is one of the most musically intriguing I have encountered in many a year. The problem of correct bass timing in speakers has been known for a very long time—decades really. But addressing it was a difficult matter in a purely analog world. In the DSP world, things are different. The details of all this will require some explanation, and, in fact, there are aspects of this device that I have not yet been able to clarify completely in technical terms. But in listening terms, the results are remarkable.
So before explaining what SAM does and why such correction is necessary, let me describe in summary what the sonic effects are before I go into the underlying technical considerations.
How SAM Sounded on Speaker X
I tried the SAM system on a medium-sized stand-mounted speaker of high sonic excellence. Which one, you may ask? Pardon me if I do not say. In my many writings about DSP room and speaker correction, I have found that any speaker I describe as benefiting from these adjustments is almost automatically devalued. The immediate effect on many readers is: “So that particular speaker needed help, but that is because it is not such a good speaker. My marvelous speakers do not need help like that.” This is always—or almost always—wrong. Your marvelous speakers do need help. Virtually all speakers do. And I definitely do not want to let anyone think that the fact that the SAM system improved the particular speakers I used as criticism of those speakers. In fact, the medium-sized speaker I am describing is one of the great speaker designs of the world. Reputable reviews from other people have called it “one of the best,” and so it is. Call it Speaker X from here on.
Speaker X is almost flawless from the mids on up. But because of its medium size, it lacks deep bass extension. And because it lacks extension, it also lacks a certain precision in what bass it does have. I shall explain later why this is inevitable. No passive speaker that lacks full bass extension—and even more than nominal 20Hz extension is needed—can do bass timing correctly. This is just a fact, even if it is an unfamiliar one for people not acquainted with filter theory.
On to the sound with SAM: I have been enjoying listening to the Telarc Ravel-Borodin-Bizet CD. This is attractive music, which bears repeated listening, and it is very well recorded. But the music is not so profound that one is reluctant to listen to it over and over in quick succession and with attention in good part to the sonics. (One really should not use Beethoven’s Missa Solemnis as an audio-test item.) On Speaker X, it presented a quite convincing picture of an orchestra, something that I did not feel foolish listening to coming from my Wednesday rehearsals with a large orchestra (Santa Monica Symphony). But the bass was somewhat limited and not entirely precisely defined. Things such as the low string pizzicato passages were inclined to be a bit blurry compared to reality. And the blats of trombones and the like were slightly less blatty than they should have been.
Enter the SAM system. All of a sudden—no ifs, ands, or buts—the bass acquired much greater perceived extension and a lot more additional perceived precision. Trombones blatted as they should, when appropriate. Low string pizzicato passages acquired the resolution of real life. One could hear what went on in the lower frequencies much better than before. Indeed, the whole sonic picture was cleaned up. The result was a sound that was far closer to a real orchestra—among the top echelon in reproducing orchestral sound of any audio systems in my experience (within reasonable dynamic limits).
Let me be careful here. Speaker X is a medium-sized speaker. It is not going to be able to pump out the huge dynamic levels of the Carver ALSes (a large line source with a high-powered subwoofer) or the Cerwin-Vega CLS-215s, with their two 15" woofers per side. Speaker X did acquire enhanced dynamic range, playing louder than before without audible distortion, but no electronic pre-processing or active management could make a speaker of this size and nature fill an auditorium. In a domestic living room, however, volume was more than sufficient, and accuracy was superb.
Audio reviewers are an excitable lot on the whole. But my long-term readers are well aware that I am more along the lines of “the ice-cold bastard of audio reviewing,” to borrow Janos Starker’s phrase. (I once congratulated him after a deeply romantic performance of Dohnányi’s Konzertstück. He said thanks and then remarked with a wry grin: “That ought to show those people who call me the ice-cold bastard of the cello.”) Almost everything calls out my critical instincts rather than wild and uncontrollable enthusiasm. The SAM system is an exception. This is really something extraordinary and revolutionary, and I don’t mind saying so. It is musical magic. But it is not literally magic—no magic signs, no mandrake roots dug up at the dark of the moon and dried pointing exactly north. This is a serious technical design that works in an explicable if somewhat complex way. What the device does is measurable, as I shall get to later. (At least I could verify part of it in measurements. Other claims made by Devialet seemed correct in listening terms, but I could not check them all with the measuring methods I had available.)