The Lifelong Quest For Musical Realism:
Spectral Audio's Rick Fryer & Keith Johnson Talk With Robert Harley
Reprinted from Fi Magazine - June 1998 Issue
Spectral Audio occupies a unique position in the high-performance audio industry. The company isn't obsessed by the need for growth, to introduce products at every "price point", or to achieve maximum return on investment. Instead, Spectral chooses to build its products in limited quantities, to "push the envelope" on every new design without regard for development cost, and to run the company almost as an independently-funded research laboratory rather than as a for-profit enterprise.
Founded by Richard Fryer in 1975, Spectral benefits from the design talents of Keith O. Johnson. Keith is not only one of the greatest classical-music recording engineers of all time, but an outstanding analog circuit designer, digital circuit designer, test instrumentation inventor, and magnetic tape expert (one of his innovations in the early 1960s made high-speed cassette duplication possible). Keith is also the co-inventor of High Definition Compatible Digital (HDCD), the process for improving the sound of the compact disc. Discussing music reproduction with Keith reveals his astonishing depth and breadth of knowledge of how things work, as well as his deep, lifelong commitment to recreating the original musical experience. It's worth noting that at age 14, Keith built a record-cutting machine from scratch that used pieces of cardboard sprayed with shellac as the recording medium.
To find out more about the philosophy behind Spectral Audio, I sat down with Richard Fryer and Keith Johnson. I began by asking Richard what led him to start Spectral.
Richard Fryer: My interest in audio began many years before starting Spectral, as a young person building music systems through grade school, high school, and into college. I also did some reviewing, and worked around high-end companies for many years. My area of study in college was psychology - cognitive systems in particular - and developing methodologies for research in psychology. Out of that came and interest in perceptual studies, mostly to complement my interest in music playback.
High-end has always been an American tradition. Spectral really continues on the shoulders of those companies hat have come before: Marantz, Scott, Fischer. We've availed ourselves of concepts and component technologies that allow us to address some fairly fundamental issues in music-reproduction systems. That in a nutshell is what Spectral is about.
Tell me about your approach to product conception and design.For practical reasons, component technologies and design philosophies found in high technology fields are not pursued in consumer audio products. Because we had backgrounds in computer design and instrumentation design, it was obvious to us that there were not products of this type on the specialty audio market. We knew what some of the advantages would be if we designed to a higher standard.
We continue to take our lead from what is done in the more ambitious fields, where state-of-the-art is absolutely required. Solutions in those technologies have evolved over decades, but have not been applied to the challenges of reproducing musical events. We're taking fundamental solutions that have been present in other design fields and applying them to new areas and new challenges. That's probably as much as anything what Spectral is about.
We're undeterred by price. We're undeterred by how long it would take to develop a product around a technology if we're drawn to its potential for superior and accurate music reproduction. That potential may be enough for us to commit to years of development time to realize an advantage. This approach has served us well in designing to a level of refinement that is relatively rare in high-end music products.
Fryer: We've approached it somewhat naively, figuring that we had some interesting solutions to classic problems in amplifier and transducer design. We didn't look at it economically. We didn't look at it in terms of the resources necessary to meet those goals. We simply said, "These are the areas in which we think we can express something new." We wanted to eliminate all other considerations and simply say, "Price is not a necessary object in pursuing the state-of-the-art."
Keith, how does your involvement in recording and designing recording equipment affect how you approach designing consumer playback equipment?One starts recording to have the thrill of making a permanent archive of a musical experience. There are many, many obstacles along the way. You hear things that don't quite sound right, that need to be improved - or in some cases done from scratch - to achieve the goal of making something that matches the experience of a live music event. It takes a long time, but it's a wonderful experience. It's the same with designing playback equipment. There is definitely some aspect of black art. But I wish I knew what makes timebased jitter, or thermal tails, or many other things so audible and yet so difficult to measure as an end result.
Keith Johnson: Listening to the microphone feed with a live orchestra on the other end is an invaluable experience. You hear certain problems - a granularity, or smearing of information, or lack of detail, or a number of other things that can go wrong. By listening to the live feed and knowing what goes on inside the electronics, you can go back to the bench and figure out what caused the problems. Usually, I have to build my own test equipment. Not only do I build microphones and the rest of the recording chain, I end up building test equipment as well.
Spectral introduces new products very infrequently. Why?We average somewhere between 24 and 36 months of development on a typical Spectral component. The first year is usually spent chasing down basic research into what the product could be potentially. We look at available technologies, at where past designs have failed, and what we hope the new product will do. Another year is necessary to build prototypes. A final year is almost always taken in laying out and tuning the design, accompanied by multiple, multiple listening sessions to finally nail the sonics and refer the product's sound back to Keith's live microphone feed. Three years go by very quickly with that type of commitment to the outcome.
The beauty of that investment in time, energy, and capital is a music component that can have many years of very active market life to add to its value and the buyer's investment. Whereas products with lesser development and refinement typically don't last in the market that long. Very refined and thoughtfully optimized components, even in our high-end industry, are quite rare.
Products done at that level of commitment have staying power. They continue to be competitive in some cases for a decade. A very fine audio component is not inexpensive to manifest as an instrument. It takes care. It takes time and it takes craftsmanship. This is something that you don't always get in any product, but you should require in a high-end audio component. You're paying for it. You should get it. But this isn't always so, even at these elevated prices. It's a great concern to us to build a music component that, years later, you would look back on and say, "This component was honest to the musical source. There was no embarrassment. It has been a faithful servant to the recording." The highest praise that we could realize for our components is years later to have it said that these products were correct, that they were classics, that they represented honest and fundamental thinking in solving problems.
Fryer: When attempting to create a state-of-the-art component, the commitment to development time can be vast. The multiple listening sessions, with minute part changes and massaging, can add weeks, months, and even years to the design. But they're essential for the ultimate refinement
The SDR-2000 (Spectral's digital processor) comes to mind.
Johnson: The basic design goal of the 2000 was to make a converter with one part per million accuracy with respect to non-harmonic artifacts. That machine comes very close to that level of performance. The critical parts can't be any better. The machine continues to be the statement of the best we can do until more fundamental improvements come about.
Fryer: In terms of digital products, our philosophy has always been that as a front-end to the Spectral system, nothing less than the state-of-the-art is acceptable. We didn't see cost-effective or limited versions of our digital technology as being useful. We left that to others, because we felt our contribution could be the very lengthy, ground-up development programs for digital understanding and breakthrough. We didn't feel that other companies would have the interest or the ability to pursue that path.
It's not an overly profitable business. From a financial point of view, it's almost suicidal for high-end companies to attempt state-of-the-art digital design. You can'' recoup the man-years necessary to derive fundamental breakthroughs in digital science. This would normally be left to bigger companies. But those big companies look at it from a marketing point of view and also don't always see the return on investment.
It was of great personal interest to us, and an aesthetic and technological challenge, to attack digital problems. We felt strongly that digital audio introduced real limitations in fidelity to music reproduction. Yet it was a future that we were all forced to partake in. If we were going to be critical of these digital systems and products, we had no one to blame but ourselves if we didn't roll up our sleeves and make a contribution.
I think it's very clear that Keith has made a contribution that goes far beyond the high-end. It now affects the main-stream recording industry [with the Pacific Microsonics professional HDCD digital encoder - RH], and frankly, in the way that people have opened their minds to potential improvements in digital. Today, you hear mainstream recording engineers and industry giants say, "How can we have better digital playback?" This is new and fantastic because this is what the high-end has been saying all along as a voice in the wilderness. But today, we're not the only ones who believe it. When everybody agrees that it's time for change, then things start to happen. This is a very good time in terms of improvements in the state-of-the-art because it's not just the high-end that sees the need for change and for improved quality.
Johnson: We're seeing exactly the same thing at the world-class studios. People who are making the major waves in the mastering industry have the same sensitivities and same perceptions as people in the high-end. At Pacific Microsonics, we've spent a good five years dealing with A/D conversion, filters, and system characteristics that are needed to achieve an essentially artifact-free recording. Other companies are also making converters with high accuracy - not as good as the Microsonics encoder - but essentially much better than in the past.
The mastering houses are producing some very fine product now. In the past, I always felt that CD was heralding the dark ages of sound. The older conversion technologies pushed you away from the artist. Now, with everything done very well, there is no more involvement with the artist. There is more interest in music. That trickles down to much better software, which benefits everybody.
Spectral has long advocated the "system" approach to its products, in which you engineer an entire playback system that happens to be in different chassis. You stress that to realize the full potential of a single Spectral component, it needs to be used in an all-Spectral system.The advantage is that when you're able to look at complete systems, you can eliminate so many variables that affect resolution and linearity in music playback. You can optimize complete systems for that one goal of resolving the recording with no interpretation. In main-stream products it's necessary and more cost-effective to look at what the intrinsic component does. But that's because they don't have to achieve an ultimate level of performance.
In high-end you would like a system of ultimate transparency and ultimate resolution. The system should have no qualities of a musical instrument itself, and only be a reproducer. That's a luxury very few designers or companies have the opportunity to pursue. But it's something that we wanted very badly to pursue. Only when you control all the factors in a recording or a reproduction system can you begin to see what contribution the individual components actually make. Only then that you can discover what the most important things are. That may result in a system that's not for everyone, but it can result in a system of remarkable refinement.
Fryer: We didn't have the luxury of building complete systems in the early years. But with a body of development that is now about 22 years old, we have been able to - at least once - express our state-of-the-art goals in preamplifiers, power amplifiers, and digital front-ends. And with the help of Bruce Brisson at Music Interface Technology (MIT), we've essentially been able to express that in interface systems for our wideband, fast-settling circuitry. That's a luxury few companies have the opportunity to pursue or probably can afford.
Johnson: It's always essential to have a fully-integrated or fully-interfaced system. The ideal power amplifier would be one that in no way can interact with the speaker crossover. In other words, whenever you hook the amplifier to the loudspeaker, the speaker's crossover and drivers behave as if you haven't changed anything. Yet a power amplifier that's very fast is going to put out some very high frequencies. The speaker cable bust be able to damp or absorb those frequencies without causing reflections or rebound from the unknown parts that are inside the loudspeaker.
That requires a special cable. In other words, the faster the amplifier, the more one has to be very careful about the cable connecting the speaker. In a system approach, the advantage is that the cable is done right. The speaker sees virtually the zero or defined impedance that the crossover was designed with. A cheap cable might have inductance which will actually detune some of the parts within the crossover and make the crossover behave differently. A very small change of inductance in the cable can make a big difference in what's happening inside a speaker. By building all these parts as a system, the interactions are designed out and the whole thing simply works better.
Fryer: In the quest for the Holy Grail of the ultimate reproduction, there is no alternative to designing tuned systems. Anything else is a lossy and sloppy design that cannot promote ultimate fidelity. We are interested in ultimate fidelity to the recording. Anything else is the arrogance of interpretation.
Certainly in more compromised design, there is every reason to balance the compromises. But in the high-end, we should have a commitment to the ideal that says the recording should be preserved at all costs and interpreted as little as possible. This is the only way we know to achieve such an end. To do anything less is an improper use of the artist's property from our standpoint.
Although lesser systems introduce colorations and have various distortions, there should always be a goal of making the system as benign as possible. The system should have the best possibility of communicating the music to the listener. That's the beauty of our business. There occasionally can be the luxury of pursuing this on a very idealistic and artistically ethical level.
Johnson: Another factor is that we want to eliminate what I call the "one and only" system. That's where one takes a group of components and starts sifting through these to find some combination of parts that seems to work the best. Then if you change any part of the system, it suddenly doesn't work. The system becomes a complex interaction of filters. If you finally get something pretty good, you can't do anything with it. You can't change it. If one component comes out and you put something else in its place, it's not going to work right. That's not a particularly good value to a dealer wanting to sell systems, nor to a consumer because he is not able to upgrade as the state-of-the-art advances.
Why do you design such fast, wide-bandwidth components? Although the highest audio frequency is 20 kHz [20,000 Hz], your products go out to 3 megahertz [3,000,000 Hz].High-speed circuits give you the ability to hear inter-transient information. With trumpets, for example, you get a brilliant sound where you can almost hear the individual cycles of a trumpet. That quality is related to interplay between various types of distortion, settling time, and energy storage within the feedback loops. If those factors aren't under control, the system will sound muddy and murky. If one tries to brighten it up to compensate, then one has harshness without much redeeming musical information.
Johnson: The Spectral philosophy has been to build products that are very fast and have very fast settling. Very fast capability means high bandwidths. High speed isn't important just to go from point A to point B quickly, but in preventing the circuit from having a memory that the musical even happened. Now I have a fresh start for the next event. That's speed. The electronics must settle down instantly from the loudest signal to the smallest signal, which is a ration of far more than a million to one. And they must do this within about ten millionths of a second.
Explain to us how an audio circuit changes its characteristics based on the nature of the musical signal it has just processed.A similar phenomenon happens with magnetic fields, particularly in power amplifiers. Even though the circuit inside might be very fast, the end result is not speed because you have to wait for the event. The feedback and compensation parts inside amplifiers store charge in a non-linear way when hit with a transient event. The charge then bleeds off after the event, creating a settling problem. You may not hear it as a sound, but you perceive that the sound is not particularly clear.
High speed isn't to get from point A to B fast. It's that you don't want anything to remain once you've gotten there.
Johnson: A common problem in circuits is "thermal tails." If you shock the circuit with a transient audio event, then the internal parts are stressed. This stress creates heat which remains after the event, which in turn creates a memory in the circuit. It takes a finite amount of time for things to cool off. When the next transient comes along, the circuit behaves differently because it is warmer. This problem is quite audible. The faster the circuit, the less energy stored, and the less the circuit changes from transients that have just occurred.
What are some musical examples of the problems you're describing?The other factor you get with speed is lightness. The music can have a tremendous amount of high frequency energy, and yet you don't feel assaulted by the energy. It can be very bright, yet not at all harsh or hard to listen to - exactly the way a live instrument is. If you tried to reproduce a trumped at a live volume, you couldn't stand to be in the room with the loudspeaker unless you get the speed issues right. With high-speed circuits, the harshness goes away and the bloated character is gone. It sounds lighter.
Another very good example is pace and timing. If the energy settling is relatively slow, the sense of pace is slow. This is one of the things that the British people have been listening to for quite awhile. With fast circuits, it sounds like the tempo is faster, with more musical activity.
Johnson: The faster the transient-settling time in a system, the more easily you can reach in and pick out tiny details in an otherwise very big sound. For some reason - and I don't know exactly why - the staging usually appears wider when the speed and the settling is better. That seems to give a bigger perceptual window to reach in and hear fine detail. In other words, if the stage is small, then everything sounds cluttered in a small space.
Fryer: Years of research have gone into practical ways to turn this ideal into products. We've worked on it for 22 years now. It's taken us over two decades to address these issues of speed and settling and resolution, first in preamplifiers, then power amplifiers, then the interfaces between them, then the front-end. Bruce Brisson has addressed these problems with us in his cable designs.
Now we've gone on to very fast, quick powerful loudspeakers that are capable of very complete settling and ideal behavior. We're working with some speaker manufacturers to achieve these same ends in the mechanical systems, but it takes years.
These techniques of fast settling and terminated systems come from the world of radio, communications, antennas, microwave, radar - all of these fields. These are almost mature technologies. But we can draw inspiration from their solutions and apply them to our low-frequency realm of audio.
The question has always been from outsiders "Can you demonstrate that these super-fast circuits really benefit what happens in the audioband?" The history of our company has been emphatically yes, you hear it. You don't have to do a whole lot more than compare the reproduction with live music, or compare the microphone feed to the final playback, to see the tremendous necessity of designing products this way. The losses are so sever from the front of the recording system to the playback, but we all accept it as high fidelity.
We have nothing to lose by applying these techniques, and everything to gain. Because on some levels the reproduction system is still so crude, so non-linear, and resolves only a part of what the event was about. That's why it's potentially a lifelong quest.
All files are copyright 2008 by Spectral Audio, Inc.