The history of audio is littered with the skeletons of “breakthrough” signal-processing technologies that reportedly improved the listening experience. From the “Hafler Hook-up” of the 1960s to today’s sophisticated DSP-based systems, there’s been no shortage of attempts to enhance music listening via manipulation of playback signals.
The problem is that none of these systems, at least in my experience, have worked. They end up sounding contrived and hokey. Rather than bringing us closer to the absolute sound, these technologies take us further away by creating “effects” that are supposed to mimic the live experience but end up calling attention to themselves as effects. I’m not talking about the absurd DSP programs on AV receivers such as “Concert Hall” or “Jazz Club” that attempt to synthesize acoustic reflections, but rather about the long list of “spatial enhancers” and other “psychoacoustically” inspired analog-domain technologies, usually involving sum-and-difference signals, delays, and filters.
I thus regarded with tremendous skepticism the claims of BSG Technologies for its QOL (rhymes with “coal”) technology incorporated into a device called a “Signal Completion Stage.” QOL is a playback process that reportedly restores the natural phase relationships in audio signals, resulting in improved sound. The company behind QOL, BSG Technologies, plans to license QOL to hardware manufacturers in a variety of fields including car audio, cell phones, telecommunications, and motion pictures.
To demonstrate QOL, BSG Technologies developed a $3995 stand-alone box (the “Signal Completion Stage”), the subject of this review. Although this box allows evaluation of the technology, the optimum implementation is within a preamplifier, DAC, or other component. QOL decoding is so simple that an analog implementation can be realized with about three square inches of circuit-board real estate, and in the digital domain with a very small amount of processing power running on a chip. Given the development time of high-end audio products, it will likely be some months or years before QOL is available in products from a variety of high-end manufacturers—if it ever is.
The device I received for review is fairly elaborate, offering four analog inputs on both RCA and XLR jacks, dual pairs of outputs on RCA and XLR jacks, fully balanced operation from input to output, source-switching, a mono button, and a remote control. A “Bypass” button allows you to switch QOL decoding in and out of the circuit, which is convenient for listening comparisons. The Signal Completion Stage integrates into your system either between your source components and the preamplifier (thus the multiple inputs), or between the preamplifier and the power amplifier. The device is pure analog, with no analog-to-digital or digital-to-analog conversion. BSG Technologies sells the device directly to consumers and offers a 30-day money-back guarantee.
Despite my skepticism, I listened to an early prototype nearly two years ago and was encouraged by what I heard (see my report in Issue 201). BSG Technologies is quick to stress that QOL isn’t a signal-processing technology, in that it doesn’t add anything to the signal that’s not already present. Moreover, it differs from the myriad techniques used in the past (extensively referenced in the QOL patent application’s “prior art” citations), most of which involve extracting the difference channel from a stereo signal (left minus right), equalizing, delaying, filtering, inverting, or performing other psychoacoustically based manipulations of that difference signal, and then mixing it back into the L and R channels. The BSG Technologies Web site offers this explanation: “Instead of ‘adding’ a host of processing techniques intended to create ‘effects,’ we have simply found a way to extract information already present in the recordings, but otherwise hidden in conventional reproduction.”
The patent-application abstract offers a more detailed description: “The present invention relates to an apparatus and method for redeeming otherwise closed and concealed information contained in audio signals. This includes both the primary reference signal, and a plurality of redundant duplicate signals, substantially identical except in relation to magnitude and phase, for the purpose of unfolding, or opening the audio signal content into layers that result in an omni-directional acoustic signal, representing the sound as it would behave in nature. The audio reproduction system uses an in-phase circuit and a separate phase layering technique circuit to drive independent multiple mixed channels to produce an open, substantially complete sound from a discrete audio signal, for the purpose of enabling a substantially complete audio signal to be formed, or to transform existing incomplete audio into a substantially complete audio signal.”
The key term is “phase layering.” In fact, the patent application’s summary refers to the invention as a “phase layering apparatus.” I’ll try to summarize how the device works based on my reading of the patent application. The input signal (which comes from one of your sources) is used as the phase reference. This signal is split into multiple copies of itself, with each copy independently phase-shifted by varying amounts (the “phase layering”). The multiple signal components of varying phase shift are then combined into a single signal along with the unaltered reference. This single signal thus contains the original unaltered signal and multiple copies of itself that have been subjected to varying amounts of phase shift. Note that this acoustical “decoding” can be performed in an active circuit, or even passively in a loudspeaker that has multiple voice coils. The patent application states: “By applying these angles, of 45°, more or less, layers of phase form into a final, complete audio signal composite that provides a virtually spherical acoustical signal.” This is an extremely simplified explanation; those who are technically inclined are invited to read the patent application.
I’m not sure, however, how a single, fixed process can acoustically “decode” the universe of diverse audio signals. Various miking techniques—XY, M-S, spaced omnis, spaced cardioids, ORTF, pan-potted multi-mono, etc.—capture the soundfield in very different ways. Moreover, these techniques are often augmented with accent mikes on specific instruments, not to mention the use of dummy heads, spheres, and baffles, or the varying sensitivity patterns of omnidirectional, cardioid, hyper-cardioid, and figure-8 microphones, and the relationships between the microphones, the instrument, and the surrounding acoustic. It’s obvious that some fundamental loss occurs in the recording process, but not so obvious if that loss involves out-of-phase signal components that can be restored to their original condition, or how a single phase-“correction” algorithm works for all recordings.
Nonetheless, the proof is always in the listening.
I installed the BSG Signal Completion Stage in my system initially between the preamplifier and power amplifier using the balanced inputs and outputs. Before listening to the QOL technology, I confirmed that the box had little effect on the sound of the system when in “Bypass” mode (no QOL).
Sitting in my listening chair, I could switch QOL on and off via the remote control. And switch I did, hundreds and hundreds of times, with CD sources, music servers, SACD, and LP playback. I switched between the bypass mode and QOL so many times not because the change was subtle and hard to detect, but because it was so profound. Frankly, I had a hard time believing that the radical improvement in the sound was real, and more importantly, represented a step toward the experience of hearing live music rather than just a lateral (if appealing) change in the signal. It also occurred to me that QOL might simply be introducing a euphonic distortion. Whatever the case, there was no question that music listening was more engaging and enjoyable with QOL.
Pushing the remote control’s Bypass button to engage QOL rendered a significant change in the sound in several different areas. The single best description of QOL is that it caused the sound to “open up,” both spatially and in timbre. Instrumental and vocal timbres became brighter, but not in the sense of a frequency-response change (the tonal balance was unaffected), but in the impression of the timbres becoming “illuminated from within” (to use Jonathan Valin’s descriptive phrase of ARC products). QOL seemed to strip away a layer of opacity, allowing the instrument’s tone colors to become more vivid and alive. Removing this opacity wasn’t the cliché of “lifting of veils,” in which the listener has the impression of a fine scrim being removed between him and the music, but rather that the opacity heard without QOL was imbedded within the instrumental timbre itself. QOL widened the palette of tone colors, the density of those colors, their vividness, and their life. This richness of timbre and vividness was not an artificial “Technicolor” rendering that sounded hyped or artificial, but rather revealed the inner brilliance of instruments and voices—a heightened sense of chiaroscuro. With QOL engaged, Joe Morello’s superbly recorded cymbals and hi-hat on Morello Standard Time [DMP] sounded brighter, but paradoxically, not because of more treble energy. Rather, the sound had a more open and vivid quality that sounded “lighter.” Woodwinds and brass instruments sounded more like they do in life, with greater presence and immediacy, but, again, without sounding artificially forward or pushy.
The change in timbre represented about half of what QOL can do. The other half was in the expansion of the soundstage in every dimension. Hitting the remote control’s “Bypass” button to engage QOL was almost equivalent to looking at a landscape with one eye and then opening the other eye and seeing the landscape suddenly pop from two dimensions to three. The soundstage not only became wider and deeper, but the most significant difference was in the sense of individual images existing independently from other images. That is, the soundstage became less homogenized, congealed, and closed in. The sense of air and distance between images was suddenly palpable. Instrumentals and vocals were no longer “stuck” to each other, instead expanding outward, forward, and backward. QOL also fostered the impression of greater soundstage height, as though a lid had been lifted. The sense of top-octave air riding above the presentation was palpable with QOL.
Turning off QOL collapsed the sound toward the center, tended to make the soundstage congealed and homogenized by comparison, and made timbres somewhat “hooded” and veiled. The sweet spot was also narrower without QOL. Try listening without QOL and swing your torso left and right in the listening seat; you’ll hear the image shift toward the direction of movement as you hear more sound from the closer speaker. Repeat with QOL engaged and the soundstage stays solid until you get much farther away from the center-line. Dynamic reproduction was also improved with QOL. Drums and percussion had more realistic transient detail, both in the steepness of attack and in the length of their decay. The increased speed at which transients seemed to reach their full expression increased the “jump factor.”
The degree of improvement rendered by QOL varied with the recording. Some LPs, CDs, high-res files, and SACDs were somewhat better sounding with QOL; with others the difference was nothing short of dramatic. Paul Simon’s Graceland is a familiar album that most of us own, and it’s one that particularly benefits from QOL. Listen to the unaccompanied voices that begin “Diamonds on the Soles of Her Shoes” and hear how the timbre of the voices becomes more present and realistic, and how you can better hear the individual singers. Listen also to how the soundstage widens and deepens. Or how Simon’s centrally placed vocal seems to hang in space as a separate element rather than as another sound within a homogenous fabric. His voice also moves back slightly in the soundstage and becomes a little more diffuse and defocused.
To see if there was any correlation between the degree of change in the sound and the recording technique, I listened to A Meeting by the River on the Water Lily Acoustics label reissued on SACD by Analogue Productions, which is a pure Blumlein recording (a pair of figure-eight microphones crossed at 90 degrees), a recording I made using an X-Y pair (crossed cardioids), and a pan-potted multitrack mix I’d engineered. The recordings made with coincident microphone techniques (Blumlein and X-Y) seemed to be less affected spatially by QOL.
I realize that the BSG Technologies Signal Completion Stage will be controversial, and that I’m going out on a limb with my enthusiasm for it. Frankly, it seems too good to be true. But try as I might to discover something about the sound that was antithetical to high-end values, I couldn’t. From my understanding of the technology, however, it is difficult to see how QOL makes the acoustic output from your loudspeakers more like that of the original soundfield. Is it appealing? Undoubtedly. But is it more accurate? I’m not sure—it’s entirely possible that QOL is a euphonic distortion. It’s not euphonic in a way that “tubey” electronics are; rather, QOL produces a sound that I think is closer to that of live music, even if the means to that end may not entirely be in accordance with fidelity to the original sound.
Only a few people so far have heard the Signal Completion Stage in their own systems, but I’ve had conversations and e-mail exchanges with one of them, a 40-year industry veteran, whose reaction was identical to mine right down the line, from the initial skepticism, to our specific listening impressions, to the desire to keep the Signal Completion Stage in our respective systems. I’m still trying to come to grips with QOL, and am not entirely convinced by the technical explanation or that it represents greater fidelity to the source, but I can tell you this; every time I sit down to enjoy music now, I’m listening with QOL.
SPECS & PRICING
Frequency response: 10Hz–50kHz +/-0.25dB
Signal-to-noise ratio: >106dB
Input impedance: 20k ohms (balanced)
Output impedance: 50 ohms (balanced)
Inputs: Four balanced on XLR jacks, four unbalanced on RCA jacks
Outputs: Two balanced on XLR jacks, two unbalanced on RCA jacks
Front-panel controls: Input select, bypass, mono, power
Dimensions: 17" x 3.3" x 13"
Weight: 25 lbs. (shipping)
3007 Washington Blvd., Suite 225
Marina Del Rey, CA 90292
Outside the Box
QOL Inventor Barry Stephen Goldfarb Talks with Robert Harley
Barry Stephen Goldfarb is the prototypical inventor who seeks novel solutions to seemingly intractable problems. Goldfarb became self-educated in music, science, technology, and art. In addition to teaching audio engineering and acoustics at the university level, Goldfarb, alone or with others, has been awarded more than 50 patents in acoustics and audio electronics. He is the principal of BSG Laboratory and the inventor of the technologies marketed by BSG Technologies, LLC.
Goldfarb rarely leaves his Florida lab, but recently visited Southern California to oversee the installation of his QOL (rhymes with “coal,” and represented by the trademarked acronym “qøl”) technology at the Segerstrom Center for the Arts in Costa Mesa, California. I sat down with him to get some insight into the man and how QOL works.
Robert Harley: Tell me about your background and what led you to create QOL.
Barry Stephen Goldfarb: My background is more on the artistic side. Specifically, I played music at a very young age and grew up in a creative environment. I became a professional musician and actually used music to support my greater goals in art. That gave me the time, the money, and the freedom to do the things I wanted to do and still remain creative.
I’ve been working on a project all my life—a multi-sensory project. Think of it this way: If music is fundamentally organized sound, what if we could organize light, and the molecules that enable us to smell, and structure all these different variables to build a whole system out of these ethereal, ephemeral elements that are really in a sense non-material materials? I’ve spent my life working toward building a place made out of light and sound and color and other sensory experience.
Sound for this project became extremely important from the standpoint of creating a reference. If the reference that I was going to work with was off, then everything else would be off. When creating music and then recording music, I noticed dramatic losses between the sound of an acoustic instrument and the recording of that instrument. This was particularly true of a wonderful pipe organ. After hearing that pipe organ played back through any kind of loudspeaker, it just didn’t sound the same.
The overwhelming, all-encompassing sounds that fill space were gone, and the space-and-time relationship was totally lost. I found a great discrepancy between the real world of acoustics and the played-backed world of audio electronics. I began a search as to what the heck was missing.
My goal was to find out what was missing and see if I could correct it. I didn’t know what my limitations were, because I didn’t come at this from an acoustician’s standpoint or an engineering standpoint, or even a science standpoint. I came at it from just listening and being a musician and knowing what things should sound like.
RH: How did you get the insight that led to QOL.
BSG: I was not afraid to take things apart and fiddle with them, particularly loudspeakers. I love loudspeakers. They’re like living things to me. Everything else is just stuff. But a loudspeaker is the heart and soul. The revelation that led me to QOL was building an automobile audio system with 68 speakers and trying to get this all-engulfing, all-encompassing sound. I wasn’t trying to make it big and loud and powerful—in fact, quite the opposite. It was going to be quite subtle. The technician helping me accidentally wired one of the speakers out of phase and I said, “Wait, I like this better—it’s not supposed to be [this way], but I like it better.”
So I end up taking two loudspeakers on the left channel—that’s where I began—and two separate amplifiers, and I put one loudspeaker in phase and one loudspeaker out of phase, and power balanced the differences so that, to my ear, it sounded just right. I was getting the in-phase and the out-of-phase signals simultaneously, and voîlà! For me, sort of a whole color palette came on, and I was getting the sound, the tone, the color that I was looking for.
I was also beginning to get some very unusual radiation patterns. Now, of course, you have so much cancellation, not only in the out-of-phase loudspeaker, particularly the lower-frequency cancellation, but you’re also getting a wonderful radiation pattern.
I was really creating a dipole using two separate drivers rather than one driver reflecting. I extended that to every point in the car. That’s why I ended up with 68 loudspeakers. Each one radiated differently. In the world of acoustics and music, sound radiates omnidirectionally, but why doesn’t the loudspeaker do that? It’s got mechanical and electronic limitations. The crossovers and filters change the phase, which changes the time and the space, and on and on and on.
I got into the idea that what we need is a single loudspeaker that would produce all the frequencies and it would radiate omni-directionally. Of course, I didn’t know enough to know that that was impossible. Not to sound boastful or anything, but I built one, and that’s my pride and joy. And it works, and it sounds absolutely real.
The idea of accurate tonal color is based in multiple areas. You have to be able to radiate in all directions. You have to able to extract all of the spatial and temporal information that is locked in the signal and restore it. And I think I figured it out.
RH: How long did it take from that initial insight until you had an actual circuit that worked?
BSG: About seven or eight years.
RH: Can you explain the theory behind QOL?
BSG: Essentially the idea was to get out of a single signal both the in-phase information and the out-of-phase information, the way I was getting it out of an omnidirectional loudspeaker.
Now, how do you do that in a signal? I just began experimenting. Of course, if you have complete phase reversal of a signal that’s equal in amplitude and frequency you have complete cancellation. I wondered if that energy really disappeared, or if was being smothered out. In fact, my partner in this, who also helps me with my patents, Rob Clark, is an expert in this field [Dr. Robert L. Clark is Professor and Dean of Engineering and Applied Sciences at the University of Rochester—Ed.]. He’s written a book on active and adaptive noise controls, which deals with active noise filters.
But I began to try to figure out a way of tricking the signal so that part of it would play and another part might be cancelled. I then tried layering different frequency paths. Let’s say I took a limited frequency band up to, say, 3kHz. I’d let that play. Then I would take another band-limited signal from 3kHz to 6kHz and put it in the opposite phase. Now they’re playing together. They’re not interfering because the two are not really playing the same frequency simultaneously, If you keep doing that with other frequency bands, it’s like weaving frequencies. A group of frequencies will be in-phase to a limited bandwidth; another group of a different bandwidth will be out-of-phase; and I would add these layers until the entire audio bandwidth from 20Hz to 20kHz was covered. That technique produces a whole audio signal.
RH: Certain frequency bands have some inverted polarity components added back into the signal?
BSG: We call it in our patent “Phase Layering.” The number of layers could be infinite. In theory it has a minimum of two and a maximum of infinity. But the most important point was that I was trying to get out what was already in a signal.
There’s a great discrepancy between the way sound operates in the real world and the way the audio industry is building equipment. My number one goal was understanding what was locked inside an audio signal. “What was getting lost? Can we find that information or has it really been cancelled out?” It seemed to me that the information is in the signal but is dramatically lost when it is being retrieved.
RH: Is the correction algorithm the same, no matter how the signal was recorded? That is, is one set of frequency bands and phase shifts ideal for simply miked recordings as well as for multitracked recordings?
BSG: Yes, and let me answer you in two ways. I started working with mono because I didn’t want the confusion of multichannel signals. The primary work was done with a single-microphone recording, played back through one loudspeaker. Stereo was another issue. While stereo is a wonderful window into a virtual world, it doesn’t exist in nature. We don’t hear in stereo. We don’t have phantom images in life. It doesn’t require two birds to hear the bird in the middle. We hear thousands of points in space. But with stereo blending, the circuit is a little different; however, it’s still the same algorithm.
RH: How simple or complex is the decoding circuitry?
BSG: Actually, ridiculously simple once we get it down. In the patent application we have examples that show how many layers you can have. One is an electronic circuit with perhaps six layers. One is a simple passive version.
RH: Where in the signal path from the actual acoustic event to what the listener hears in the reproduction is that phase information hidden?
BSG: Let’s go back to that concert hall, and let’s take a look at the way most recordings and most audio equipment works. If we make a recording in the back of a concert hall, what we hear sounds great, but when we play the recording at home it sounds terrible—it’s too wet. What happened?
Your ears and the microphones are in the same positions, but the microphone isn’t connected to the brain. There’s a whole process that occurs between the radiation pattern and its positioning in space and in time relative to where we are. It’s just an amazing mechanism.
The microphone is going to simply pick up the loudest elements that make up sound. If the loudest element is the loudest in terms of amplitude, it will pick up the loudest amplitude thing.
RH: It’s purely pressure activated.
BSG: It’s pressure activated. I’ve taken recordings that were really wet and horrible, and I went to the other side [the out-of-phase information] and the signal was, in fact, dry. That information is also in there. That’s not something you could do with QOL now because the ratio [of direct-to-phase-layered components] is set up for 99 percent of recordings out there.
RH: Is there an advantage to digital implementation of QOL rather than analog implementation as in the box that I have?
BSG: There’s no advantage to me whatsoever. I am not a digital fan, other than its elegance and speed. But the digital implementation will be our biggest market because the world is running on digital.
RH: So presumably the circuitry can be small and be integrated into, say, a preamplifier or a digital-to-analog converter?
BSG: Absolutely. The analog version can be a very simple circuit about this big (holds two fingers apart). We have created the digital algorithm and loaded it into three different chip platforms.
RH: What applications do you see beyond music reproduction?
BSG: We have already built prototypes for cell phones and the human voice. The voice is much more natural. I see QOL going where any audio signal is being used: AM, FM, speech and voice dialogue, motion picture theaters, music, musical instruments.
RH: Why didn’t someone think of this before? Why did you think of this in the 21st century?
BSG: I think it’s because I didn’t know enough to know what I don’t know. I didn’t come at it from the standpoint of an engineer. And I think musicians, for the most part, find that serious world of engineering very intimidating. I was fearless! Fearless, in the sense that I’m a border-breaker by nature, and I’m looking for something. I wanted—I demanded—that I would be able to get through this larger project [the multi-sensory art installation mentioned earlier], and the sound was such an important aspect of it.
It’s an acoustic process. It’s not your typical processor because we’re not adding or subtracting anything. We’re relating the principles of acoustics to the way the brain would interpret that information as real once it exists in the air. And no one had done that before.