Jorma Salmi, 1948-2018

The distinguished Finnish loudspeaker designer and one of the greatest theoreticians of speaker acoustics of modern times, Jorma Salmi passed away suddenly on May 27, 2018. Salmi’s work at Gradient became a portal into what ought to be the future of loudspeaker design. After working at Nokia and as the head of the acoustics laboratory of Lohja Corporation (a major Finnish company), Salmi founded Gradient with Jouko Alanko and Mikko Paloranta in 1984. 

Jorma and I became friends over the years, and later on perhaps I shall share some personal reminiscences. But to begin with I would like to talk about his loudspeaker design work in some detail. Jorma Salmi was a man who believed in scientific truth. I think he would have wanted his life’s work to be discussed in such terms rather than in generalized rhetoric. The company’s history is nicely presented on the Gradient site so in this piece I shall offer comments on my own and other people’s reactions to his work.

The Gradient Model 1.3 and the SW-63 Subwoofer for the Quad ESL-63 

A research project that Salmi conducted with his assistant at Lohja, Anders Weckstrom, motivated the founding of Gradient. The project, entitled “Listening rooms influence on loudspeaker sound quality and ways of minimizing it,” was presented at the AES Convention in Montreux in 1982 and involved a “bypass” test in which the output of a loudspeaker was recorded anechoically and compared to the original input signal. Surprisingly, it was found that the acoustic output recorded was essentially indistinguishable—audibly so—from the input if the speaker had flat frequency response. The contrast between this result and the fact that flat anechoic response is very far from guaranteeing any particular predetermined sound in actual listening rooms led Salmi to conclude that speaker design properly ought to concentrate not just on flat anechoic response but on how a speaker interacts with the listening room around it. And so Gradient was founded to pursue a specific approach based on those learnings. (More details on the project can be found here or in the original AES preprint, which is available online.)

The first Gradient design, the Gradient 1.0, attempted to achieve the lofty goals of flat frequency response and listening-room compatibility in startling and very effective ways. The speaker had a floor-firing woofer, a large midrange driver operating as a dipole and tilted upwards somewhat to minimize floor bounce, and a tweeter line array mounted vertically. The 1.0 truly minimized early reflections. (Incidentally, a slightly revised version, the Gradient 1.3, was the first Gradient speaker to reach the U.S, and I reviewed it for TAS in Issue 80 in 1992. This review appears online here, alongside follow-up comments from Jorma and me.) Even today, it is startling to look at its in-room impulse response—seen in the aforementioned review. The impulse arrives and then there is essentially nothing in the way of early reflections. The sonic picture presented is very much as in Jorma’s original motivating anechoic experiment: The room around the speaker is taken out of the picture for a long time interval. Moreover, the room sound is down in level but also is very nearly uniform in frequency balance: The speaker has high but nearly uniform directivity. It also sounds very much like its input and it is remarkably indifferent to listening room acoustics. I subtitled my review “An exploration at the frontiers of acoustics,” and that was indeed what this speaker was; I really admired it both technically and musically, and I employed a pair in my own system at home for a long time—and on occasion I still do.

In  1991, in cooperation with Quad, Gradient introduced the SW-63, a dipole dynamic-driver subwoofer that fit under the Quad ESL-63s and also functioned as a stand for those speakers. This created a remarkable combination, as the subwoofer integrated perfectly with the dipole electrostatic Quad 63s. Many people at the time (including me) thought this composite system was one of the best, perhaps the very best, speaker system available at the time that was compatible with domestic environments. The subwoofer sold well and, in addition to the connection with Quad, helped Gradient through a financially difficult time in the mid-1990s. 

The Gradient Revolution

In 1994, Gradient brought out the aptly named Revolution. This system had a dipole dynamic-driver bass (two drivers per side, mounted one above the other) and a concentric midrange/tweeter unit on top that used a “flow resistance” enclosure to attenuate the back wave, which proved very effective. If one listened to the Revolution system outdoors for instance, and walked around behind the speaker, the sound all but vanished above the bass. Combining this aspect with the possibility of rotating the bass relative to the mid/treble unit opened up the option of placing the speakers against the wall (with the plane of the bass unit perpendicular to the wall).

Later, Gradient released an active version of the Revolution with separately powered, adjustable bass units that could also be doubled up. This system was and remains one of the most demonstrably accurate speakers ever offered that did not include DSP room adjustment—and did not even need such adjustment in most cases.

The Revolution offered superlative musical reproduction and in-room technical performance that was literally unequaled. In Stereophile (March 1997), John Atkinson published an in-room response graph illustrating the speaker’s uncanny smoothness and flatness. I measured similar results in my own room, albeit showing a slightly rolled top end, as my room is intentionally rather acoustically “soft.” In my own experience, one can get such results with conventional speakers only with the most favorable combinations of placement and room acoustics. But the Gradient Revolution could deliver such sound in almost any room. This was truly the realization of an ideal: a speaker that would provide in ordinary rooms with no specific acoustic treatment the acoustic truth otherwise reliably available only in RFZ (reflection-free zone) or LEDE (live-end dead-end) rooms. I concluded my Revolution measurement comments supplementing my review from TAS Issue 154 as follows: “To say that this is textbook performance understates the case. It is more like rewriting the textbook as to what is possible.”

The Gradient Helsinki 1.5

The Revolution offered such exceptional performance that Salmi might well have rested on his laurels. But his questing spirit continued and he brought forth a new model: the Helsinki 1.5. While this speaker may have been less “automatically” satisfactory in its setup, it set new and higher standards in stereo presentation and more. The Helsinki 1.5, which I reviewed for TAS in Issue 189, extended the concept of reducing listening-room influence to heighten the speakers’ stereo separation to a degree that escaped (and still does escape) other transducers. If a mono signal was played and the channels wired in reverse polarity, the signal at the listening position was, above 300Hz, at least 15dB down from the signal if the speakers were wired in the same polarity. For ordinary speakers at anything but point-blank range, one is lucky to get 5dB of such cancellation anywhere in the frequency range except perhaps at the very top. This channel separation made the presentation of well-made stereo recordings almost uncanny. Once again, Salmi had taken loudspeaker design to a new level.

Along the way, Gradient also brought out some smaller loudspeakers with the the Prelude/Laura series, and the Gradient 5, which used the Revolution’s concentric driver technology and flow-resistance enclosure ideas that allowed the speakers to be placed close to the wall without sonic penalty. And there was also the Evidence floorstander, a more conventional design compared with the Revolution and Helsinki 1.5 as it employed box woofers instead of dipole bass.

All of these loudspeakers offered excellent sound even if they didn’t exceed the revolutionary nature of the big three: the Gradient 1.3, the Revolution, and the Helsinki 1.5. But the spirit of “ever onward” continues at Gradient today—since 2012 the company has been in the hands of Jorma’s son Atte Salmi, although Jorma continued on as a design consultant until his death. I look forward to encountering the Model 1.4—which combines the floor-firing woofer idea with a concentric driver and spherical enclosure—albeit with the sadness of knowing that it would be Jorma’s swansong.  

My Personal Friendship with Jorma

Jorma and I became acquainted via my reviewing of his Model 1.3. We did not see each other often—it is a long way from Los Angeles to Finland. But we carried on a lively correspondence, and Jorma would visit us sometimes on the way home from winter CES shows. We discovered a number of common interests: the scientific theory of speaker design and related matters of acoustics (of course), as well as our shared interest in music, sometimes of not-so-common kinds (we were both great admirers of Gabriel Fauré for example). And then there were dogs. We both loved dogs in a way rather beyond the ordinary perhaps. I do rescue work for Dobermans, and Jorma was always happy to meet my dogs. And here I discovered a great gift that Jorma had in a direction quite different from his work: Dogs loved him.

My dog Hilda (Brunhilde) was the living embodiment of the idea that Dobermans can be standoffish with strangers. She was not aggressive but she simply would not be touched or even approached by people she did not know really well and that took a long time.  People who liked dogs had visited with us for a week and never been able to approach Hilda. When Jorma came for a visit, I told him that Freja and Sieglinde (Hilda’s sisters) would accept him fast but not to expect anything from Hilda except distance. But when Jorma walked in the door, Hilda took one look and promptly walked over to him and sat down right in front of him and waited to be petted. I was totally surprised. Jorma was modest and muttered something about how dogs usually liked him. This was an understatement. The same thing happened later on with Bernice Helen, another “you can look but you can’t touch” Dobie girl.

I was corresponding with Jorma right up to the time of his death. His last (email) letter to me was on May 17, 2018, and I replied on May 20. These letters were about speaker design and measurement, but among our recent previous exchanges had been discussion of mouth tumors in dogs (one of my rescue dogs had developed one and I was seeking advice) and of music. I was looking for a modern recording of Sibelius’ Andante Festivo, one that would be as similar as could be to Sibelius’ own recording—the Andante Festivo is the only piece for which a recording of Sibelius conducting his own composition exists. Jorma suggested Osmo Vänskä’s recording with the Lahti Symphony Orchestra from the soundtrack of the film Sibelius. I knew he would know. And a beauty it is.

It is hard to take in that I shall never again be able to ask him for advice about speakers or music, and that he will never have the opportunity to ply his magic with my new dog Zahara, who is even more standoffish than Hilda and Bernie Helen. But I know Zahara would be ready to be his friend, as I have been privileged to be. I shall miss him. And I send his family my deepest condolences.

Gradient and Jorma Salmi in History

Among speaker designers, he was well respected, but few designers have completely followed his lead so far. And the audio industry and audio public has often been somewhat unwilling to take a chance on things so revolutionary. Surprisingly many people are seemingly happy enough with speakers that are fundamentally not much different from the designs of Rice and Kellogg, who invented the box “dynamic” speaker in the 1920s.

Still, there is reason for optimism. What really works tends to triumph in the end, and I think of what Beethoven legendarily said of his later strong quartets after being informed that the public did not understand them. “They are not for you but for a later age.”  Perhaps the time will come when everyone will understand why Jorma Salmi found the right way to make loudspeakers and why he deserves to be considered to be one of the greatest of speaker designers.

Let me conclude with an illustrative and true story. Some years ago, Kavi Alexander of Water Lily Acoustics brought Sergei Chernyadiev, principal cellist of the St. Petersburg Philharmonic, over to my house to hear some audio. (At the time Kavi was working with the St. Petersburg musicians on a recording project: the remarkable Water Lily Mahler 5 recording.) In my audio room, first I played a large system that was under review and involved huge planar radiator speakers. Chernyadiev was polite but noncommittal. Then I suggested he might like to listen to my smaller personal system built around the Gradient 1.3s in my living room, indicating that the sound would not be as grandiose as it had been in the other audio room. He listened for a while and then said, “Not grandiose. But good. These speakers are what an orchestra actually sounds like.” There could be no finer epitaph for a speaker designer than that: “What an orchestra actually sounds like.” RIP Jorma Salmi.