PS Audio Perfect Wave Transport and DAC II

Perfect Wave Transport.  From time to time I’ve mentioned the dearth of adjectives available to an audio reviewer who does not like to repeat himself as he grows older and wiser and more skeptical. And the more revelatory the equipment he reviews, the more difficult it is to find adequate and accurate superlatives; the more of a struggle it becomes to do justice to the sound while maintaining verbal integrity. Audio reviewing, if it’s about anything at all, is about doing justice to the sound. And it follows that the better the equipment being auditioned—the better the sound quality—the more it does justice to the music. It’s all about the music, isn’t it. We audiophiles may love our equipment a lot, but we love our music more.

The PS Audio Perfect Wave Transport presents this problem in spades. What the PWT does for the music reminds me of an experience I’ve always ranked as the most remarkable and unexpected in years of auditioning equipment for review: my first proprietary AC power cords, by LessLoss. I was of the (I thought, rational) opinion that power cords simply could not make an audible difference, let alone the level of difference enthusiastically described in the magazines. What I learned that sunny afternoon in August, 2008 was that power cords absolutely can make an audible difference, a difference worth jumping up and down about (though I remain somewhat diffident of the theories proffered by manufacturers and designers to account for that difference).

By a similar token, CD transports simply could not make a big difference, especially given a modern DAC painstakingly designed for stability and immunity to incoming jitter. I was wrong about this, as I was wrong about power cords. Although in this case the rationale accounting for audible differences is accessable to reason rather than speculation. Almost all transports and DACs are in synchronous communication—to paraphrase Tolkien, there is ‘one clock to rule them all’ and that clock is encoded in the medium and is output by the source, to which the DAC is slaved. Although the transport clock is still calling the shots, as it were, an internal DAC clock is used by the SRC (sample rate converter) to reclock the data.

Until a few months ago was I using a Bel Canto CD-2 feeding a Bel Canto DAC3.3, both powered by their extraordinary Virtual Battery Supply. This front end produced some the best sound I’d heard. Prior to the CD-2, I had been using that impeccably constructed, very expensive, forty-five pound battleship of a transport, the Accuphase DP-90. (The Sony drive in this monster alone retails for over $1000.) I saw no compatibility issue arising in using an Accuphase transport with any modern DAC. Nor had I hesitated to express a vigorous skepticism when reviewers used that technically meaningless word, synergy, as if it explained everything. There was, as far as I was concerned, no reason why a Bel Canto transport should work better than a Accuphase transport just because I happened to have a Bel Canto DAC. But it did, undeniably so. I still don’t care for the word, or any other word that evades rational application, but it was the one word that sprang to mind when I pulled the DP-90 and installed the CD-2: synergy—better articulation, better imaging, smoother highs, deeper lows, more detail. Amazing. Where was it coming from?

And so my system remained for a year or two, during which time Bel Canto released several versions of their asynchronous link, a relatively inexpensive gadget that connects a computer to a DAC and eliminats the ubiquitous CD transport altogether. All around the world audiophiles had apparently begun spinning hard disks instead of polycarbonate disks. Had I fallen behind the times, and did it make a lot of difference if I had?

In part because of my luddite-like preference for physical media rather than computer interfaces, and because I had no room on or near my equipment rack for a notebook computer, I didn’t persue an audition of the Bel Canto asynchronous link. I suppose it could have been physically managed, at least temporarily, but I was also loathe to invest a sizeable chunk of money for necessary software simply to facilitate a review. And, as I say, I didn’t think it could make that much difference. And so, what was to prove one of the most significant sonic revelations of my audiophile career was delayed by a couple of years.

Accurately transmitting “1s” and “0s” is straightforward and pretty difficult to muck up. There are no fine gradations or subtle and cummulative distortion mechanisms as in analog data; a bit is either a one or a zero. Even degraded bits are still bits. But accurate (high fidelity) digital sound is not really about binary data as such, it’s about the relationship between that data and the timing signals that access that data. (Practically speaking, it’s also about the analog buffer stage that every DAC must have. A lot of engineering acumen is typically poured into this “simple” circuit, and for very good reason.) In other words, it’s about timing jitter. Of course there may be some other, unknown factor(s) at work here—after all there is some precident that our best scientific thinkers can arrive at risably false conclusions. Nor can any theory predict every possible permutation and every possible phenomenon. As Donald Rumsfeld cautioned, “..there are also unknown unknowns.”

Where this leaves the vaunted claims of DAC designs to be ‘immune to incoming jitter’ one must wonder. What were those designers thinking when they made such reassuring and extravagant claims? In the real world, playing real music, DACs are simply not immune to incoming jitter (as of November 2013, that is). Curiously, despite all the talk about jitter and its various and farflung sources, there’s actually only one place in a dizzyingly complex electronic circuit where jitter matters—where the rubber meets the road—and that is in the asynchronous converter chip in the DAC where a series of “1s” and “0s” are converted to an analog value. The implication is clear: if jitter really were completely eliminated in upstream logic circuits prior to the converter chip, all transports would indeed sound the same. But apparently it’s not because they certainly don’t.

When I posed this vexed “immune to jitter” question to Paul McGowan, the gracious CEO of PS Audio, he responded as follows: “The problem is the word ‘immune.’ Despite our best efforts, nothing is immune. Things can be better but not perfect. There’re too many noise issues in just connecting things—in fact, optical is still the best for lowest noise but everything else about it sucks, so even that’s not an answer.”  

When the CD first hit the world of commercial music, it is generally agreed the sound was pretty wretched. I personally could not say. I was ten or fifteen years late to the CD game. But apparently the powers that be hadn’t discovered that aptly-named phenomenon, jitter; not in components, not in circuit board layout, not in power supply design, and not in cables—but it was there, a sort of diabolus ex machina. Jitter comes in many gross and subtle forms, in different spectra and different distribution patterns, and it would seem that even a little bit of the stuff can be extremely deleterious to the sound—how little is a vexed question. In my experience, jitter doesn’t jump out and bite you on the nose like harmonic, intermodulation and phase distortion can do. Rather, I tend to notice jitter in retrospect, after it is reduced (or its spectrum or distribution pattern changes). The audible difference is verbally elusive but utterly unmistakable.

I noticed it instantly when I hooked up the PS Audio Perfect Wave transport. I thought I’d gone to audiophile heaven. Whatever one’s opinion of different DAC designs and the potential effects of jitter on the analog output of digital to analog converters, the simple truth is that there is a palpable difference between digital/clock data transmitted in real time from a spinning polycarbonate disk and digital/clock data transmitted from static solid-state memory; between the superb Bel Canto CD-2 and the superb PS Audio Perfect Wave transport.

Maybe it’s worth rehearsing briefly just what jitter is (or at least what it is to the best of my understanding). Redbook CDs are made by chopping up complex analog voltages into 44,100 slices (samples) per second for each channel. Each slice is a single point on the analog waveform. This sample rate is based on Harry Nyquist’s theorm (1928) which shows that by using a sample rate twice the highest desired frequency (in this case 22,050 hertz) it is theoretically possible to exactly recreate the orginal analog waveform. (Whether the measureable frequency response of the human ear represents the full frequency spectrum that humans can sense is very much an open question.)

Under the Redbook standard, each of these samples (or words) is 16 bits in length, allowing for 65,536 (2¹⁶) possible binary values. But the integrity of the binary sequence read from a CD, or from static RAM, or from a hard disk, cannot be guaranteed if the DAC—whose job it is to translate binary words (for example, 0101 1101 1010 0100, hexidecimal 5DA4) to one of those 65,536 values (voltages)—does not know precisely when the word starts and when the next bit is arriving, or if the timing intervals between words varies. This last is the primary issue caused by jitter. In other words, if the word timing is off, even if the value of the sample is accurate, its locus on the reconstructed waveform will not be—the analog shape will be distorted.

Perhaps the most challenging way of transmitting binary data is reading it from a spinning disc. Extracting 0s and 1s from the pits on the spiral path of a CD is a remarkable feat in itself, albeit prone to all sorts of errors. I seem to remember that the Redbook standard specifies a complete duplicate set of data (?), so if a word is physically damaged or corrupted the read mechamism will attempt to read its twin. In a worst case scenario, the sample may be completely dropped (which can be audible), but typically a damaged word is reconstructed using a sophisticated error correction algorithm known as Reed-Solomon. Additionally, linear velocity must be maintained at an absolutely steady rate by constantly adjusting rotational speed. (Reminds me of how we used to wonder, back in the days of servo-controlled direct-drive turntables, if this continuously self-correcting mechanism was ever spinning at exactly 33.33…3 RPM?) Spindle hole eccentricity, material thickness variation, imperfect flatness and roundness all result in the read-head making constant radial and vertical adjustments to maintain position, distance and perpendicularity. And this series of operations has all got to be done on the fly, in real time.

In the S/PDIF (Sony-Philips Digital Inter-Face) protocol, the timing signals are combined (multiplexed) with the digital data and transmitted along a pair of wires (typically, but not necessarily, coaxial) that we know as a digital interconnect. This signal must be demultiplexed by the DAC. This is not the only way to go. Master clock, word clock, bit clock and data may be transmitted to the DAC along separate cables. Or they may travel along separate wires in an HDMI cable (the PWT offers this option). The latter two methods are known as I2S protocol. It might therefore seem logical and desireable to include the transport and DAC in one box—no need for multiplexing or proprietary cables, fewer opportunities (perhaps) to pick up jitter. While I am sure there are superb single-chassis CD players, there are reasons, technical as well as personal, for using separates.

I have posed the question to a number of people in the industry, Why do different transports sound different? The responses have varied from downright dismissals to evasions to what I found a refreshingly straightforward, We don’t really know. Theoretically, reclocking by the DAC can greatly reduce incoming jitter (although to be sure, jitter might occur in the runs and components between the Receiver chip and DAC chip). Most modern DACs reclock. Of course there’s reclocking and there’s reclocking; implimentation is critical. (A recent upgrade of my Bel Canto DAC 3.3 to 3.5, for example, included an improved oscillator (clock); the audible improvement is readily discernable.) But even with the best of DAC designs, transports still tend to sound different, subtly, or not so subtly. “It’s all,” as one audio designer wrote me, “somewhat of a mystery.” The simple conclusion appears to be that the less jitter in the signals sent to your DAC, the better the music sounds. Even if we don’t necessarily know exactly why.

The Perfect Wave Transport’s twenty pounds avoirdupois assuredly includes a wealth of sophisticated digital and analog circuits, carefully designed power supplies, and scrupulously auditioned discrete components. I am sure a lot of time and money and passion went into its development. I would love to take a technical walk through its insides, but alas, PS Audio have not published a white paper offering this fascinating information. The PWT comes in black or silver, with an impeccable, multi-coated piano-finish top panel of quarter-inch MDF over aluminum. It has an LCD touch-screen on the right front and a blue lighted standby button on the left. The screen operates the drive mechanism and can display album cover and recording notes downloaded from internet databases. (There are RS232 and Ethernet ports on the rear for internect connection to any router.) Firmware is updatable by a SD card slot on the rear panel. It has a standard IEC power connector, RCA, AES/EBU, Optical and HDMI outputs and comes with a remote control. But its primary claim to technical excellence lies in how it handles data. This is not done in real time. Data read from the CD is stored in a 64MB chunk of RAM, then output with brand new clock signals available in either S/PDIF (multiplexed) or I2S (non-multiplexed) format. I use the former since my DAC, like the vast majority, does not have I2S capability.

Memory pre-loading can be demonstrated by pressing the Eject button. The music will continue playing for up to about 30 seconds with the CD removed. The PWT falls in that latest category of digital machines collectively called memory players. Once memory is loaded, it is compared to the original data on the disc. If there is any discrepency, the drive rereads the data, memory is reloaded and another comparison is done. This ‘reread many times’ approach does not use the error correction built into the drive mechanism. I presume that after n failures—an unlikely event—the problematical word is handed to an error correction algorithm as a last resort.

The Perfect Wave Transport is the second machine I’ve had that was designed without a—to me—essential functionality: track programmability. I am disappointed by it’s exclusion, especially when implimentation is simple and straightforward (and is practically ubiquitous). I hope this lack will be corrected in a firmware update. (The PWT uses a highly configurable field programmable gate array for firmware data and logic functionality. This is far more sophisticated than static memory; an FPGA is fully programmable to any desired logic topology, including that of a microprocessor with its own command set.)

Two further quibbles. When the disc drawer is open preparatory to playing a CD or DVD, pressing Play does nothing. The Drawer button must first be pressed, then the Play button. My preference would be for the drawer to close when the Play button is pressed, the way all my previous transports have worked. Secondly, if the drawer is open and the player is powered Off, the drawer remains open. I think that powering off should automatically close the drawer, if only to protect the mechanism from dust and possible damage.

The PWT uses a DVD drive that allows high definition files to be read from DVD media. With current firmware only WAV formated files can be read, so downloaded FLAC or WMA files must be converted (a lossless procedure). As an experiment, one reviewer burnt a large set of 16/44.1 WAV files onto a DVD (almost seven times the storage capacity of a CD) and the PWT had no problem playing them. The drive is standard red laser, not blue. Although I have a couple of dual-disc releases from 2L Records of Norway that include HD BluRay versions, this is a very rare production format, probably of more significance for surround than stereo.

Mozart: Piano Concertos 20, 21, 22, 23; Annie Fischer, piano; Wolfgang Swallisch (21, 22), Sir Adrian Boult (20, 23) conducting the Philharmonia Orchestra (Seraphim 7243 5 68529 2 1) I’ve been listening to the PWT for a few months now. Beethoven string quartets and piano sonatas, Brahms’s Clarinet Quintet, Clifford Jordan’s Live At Ethell’s, Liszt’s piano concertos, lots of different stuff, and I reached a point where I didn’t figure there were any more surprises in store. Then I pulled this CD one afternoon. Although I’ve heard these performances a dozen times over the years, I’ve not been overly impressed with the playing, and I always found the sound was pretty mediocre, regardless of the cables or electronics I was using. On the PWT the sound is little short of amazing—talk about giving new life to an old recording! And with the increase in detail and fidelity came a much clearer window and a more palpable sense of Annie Fischer’s musicianship, her sensitive and powerful technique, and the clarity and drive of her vision. What a treasure! There are certain pianists who play as if what they’re doing is supremely important. When they play, they speak directly to our hearts; they are alive, in the room with us, timeless. They transport us to a rich and rarified place where prose cannot go and which poetry can only refract.

Now, it’s been said that it’s the performance and not the quality of the audio system that really matters. And it’s true, I’ve heard enchanting performances on my JVC car radio, an old friend that is definitely not high end. But a loss of accuracy, be it from narrowed frequency response or dynamic range, from analog or digital distortion or from a masking noise floor, all result in a loss of information. In the subtle realm revealed in the absence of such inaccuracies, some pianists simply have more to offer than others. Yet you may only discover in retrospect what had been missing when you finally do hear it. With the PWT, I am undoubtedly hearing a more accurate, hence a more intimate, presentation of the recorded event.

Perfect Wave DAC.  Around this point in the review I received a hefty package from Colorado, at a time when that state was experiencing record-breaking rainstorms and flooding. (PS Audio’s main building is located on a flood plain in Boulder.) Inside the double box, suspended—literally—between two sheets of high-strength clear plastic mounted in cardboard frames, was a Perfect Wave DAC MkII in a classy white satin bag to keep it clean and free from fingerprints. (Kudos to the designer who devised a packaging technique that uses a fraction of the synthetic material that would be required using extruded polystyrene foam.) I would now be able to hear the PWT with the latest version of the DAC that was intended for it. The unit is brand new and you know what that means: burn in. So, I’m off for a while. I am going to do the thing I like best about being an audiophile: I’m going to listen to music through some really fine components.

And by the way, a PS Audio AC-12 HDMI cable was included in the shipment. It is a rather sophisticated implementation of I2S, a single HDMI cable instead of the usual three or four coaxial cables. I2S is supposed to be the bee’s knees of transport/DAC interfaces having inherently lower jitter probably for a couple of reasons: the clocks and data do not have to be multiplexed, and the clocks and data do not have to be demultiplexed. Thus two signal processing circuits, two potential sources of noise and jitter, are bypassed. The cable itself is an attractive piece of work using silver conductors, Teflon insulation, triple shielding, gold-plated, machined HDMI connectors and a delightfully old-fashioned serving pattern. The AC-12 will support data (10GBps) and clock (340MHz) rates in excess of those required for digital audio transmission.

I reckon that when two audio components are developed as a pair—and auditioned as a pair by people of different walks of life, engineers, designers, technicians, audiophiles, and people who are neither—it’s a pretty safe bet there will be something special in the result. The idea that the human ear can detect a hydrogen atom sneezing may be a bit facetious but it is not really an exaggeration. Developing and evaluating gear of this quality a complex business. I succumbed to the term “synergy” when I paired Bel Canto components, but that unexpected experience may have more to do with the fact that those components were designed, auditioned, modified and developed as a system. This procedure perhaps as much as anything is where “synergy” arises.

The descriptions on the web site of the various PWD functions, quite a number of which are new to the Mk II, is an impressive catalog of technical perfectionism and user configurability. One of the most significant is a selectable mode called NativeX. NativeX is a non-upsampling process that adds a circuit in the digital signal path. This circuit stores the data in a register and accesses it using a new set of clocks; the same thing the Perfect Wave Transport does at its output. These “retiming” circuits are based on the classic PS Audio Digital Lens and both have the same goal: lowering jitter. NativeX is specified to reduce jitter to less than 1pS, regardless of the input source. Such an extraordinary specification is also an indicator of the accuracy and stability of the piezoelectric oscillators used to generate clocks in the Mk II. When I was a ham radio operator we had little cylindrical “ovens” to stabilize piezoelectric (crystal) oscillators; I’ve no idea what steps DAC designers take to achieve stability.

All digital components are loaded with logic circuits, and logic circuits are chock full of gates—electronic switches—thousands and thousands of them. The gates in the Mk I were digital; in the Mk II, they are analog. The issue with digital gates is that in the “on” position the switching junction is saturated with electrons. Turning the junction “off” is not instantaneous but is a matter of waiting for all (or most?) of the electrons to drain away. And the finite time it takes for this to happen can, and does, vary from gate to gate. This inevitably introduces timing variations—jitter. Not very much jitter perhaps, but some. Analog switches, on the other hand, do not saturate when turned “on,” rather they track the incoming voltage, so they can turn off instantaneously for all practical purposes. Little or no jitter is thus introduced. Even the relatively small change in jitter level going from digital to analog gates is reported to be clearly audible.

It’s worth noting that both the Mk I and Mk II versions of the PWD have digital volume controls, enabling one to directly drive power amplifiers, in my case Bel Canto REF1000 MkII monoblocks. Now, some people say the music sounds better with a quality preamp between the DAC and power amp; often this preamp is based on vacuum tubes and is expensive. Others say not. I actually compared these two setups in the past and found I preferred not using a separate preamp, a confirmation of the ‘less or more’ philosophy.

Controlling the volume digitally is another solution in which circuit design and execution are all-important. In the PWD, digital volume level is represented by words that are 32 bits long, so at relative volumes ≧50 there is no bit stripping. (Bit stripping is a truncation of less significant bits resulting in a loss of resolution.) As a point of reference, the relative volume levels I use typically lie between 65 and 85.

The Mk II, like the PWT, has an LCD touch-screen on the right front and a blue lighted standby button on the left. The screen displays controls for digital sampling, digital filter, absolute phase (audible to some people) and source, as well as a horizontal bar indicating relative output level and a numeric display showing percentage of maximum level. The source control cycles through USB, I2S-1, I2S-2, AES/EBU, RCA and Optical. All of these, including USB, accept inputs up to 32 bits at 192kHz. The sample-rate control cycles through Native, NativeX, 44.1 kHz, 48 kHz, 88.2 kHz, 96 kHz, 176.4 kHz and 192 kHz. Native and NativeX, as their names suggest, do not use sample rate conversion. Paul McGowan thinks SRCs deteriorate sound quality and recommends using Native and NativeX. But since sample rate conversion is all the rage, the circuit design includes a bypassable SRC stage before the very sophisticated Wolfson W8742 DAC chip.

The filter control selects one of five digital filters and has a fifth choice that automatically selects one filter for 44.1kHz CDs and another for higher sample rates. The five filters offer different pre-ringing, post-ringing, group delay and frequency-versus-phase parameters. The rear of the unit has all the input jacks, RCA and XLR outputs, an SD card slot for firmware updates and a bay for an optional Bridge card enabling connection to a home network. The Bridge card has its own Digital Lens circuit (data streamed over a network is extremely prone to jitter). There is also a rocker switch to control AC power.

The term “blacker background” has been used so often by audio reviewers it’s clichéd—how, after all, can black become blacker? Howbeit, the experience is real enough, even if the term used to describe it is imprecise. I suspect what’s going on with the ‘blacker background’ phenomenon may be a reduction of a certain type of correlated noise that most of us don’t so much hear as sense. Or perhaps this business of ‘blacker background’ is another aspect of lowered jitter, just as nuance, presence and articulation seem to be. In any case, the work that went into the design and execution of the Mk II is there for anyone to hear.

For my CD evaluations I used NativeX sampling mode and Digital Filter 1: minimum phase, apodising; characterized by low pre-ringing, minimal group delay, minimized post-ringing, good phase versus frequency, and a sharp cutoff slope. The user manual recommends Filter 1 and I have not experimented to any extent with other filters; reviewers that have describe the differences as extremely subtle. As for NativeX, it seems an obvious choice to take advantage of another level of jitter reduction. The SRC positions do sound different than Native/NativeX, and I’ve encountered articles that purport to show that sample rate conversion increases the noise floor. SRCs are definitely a vexed issue. I did not pursue this line of inquiry. And finally, I thought the I2S (HDMI) interface sounded better than XLR (Wywires Gold) and used it throughout.

Padre Soler: Six Concertos for Two Organs; Tini Mathot and Ton Koopman, pipe organs (Erato 2292-45741-2) Padre Antonio Soler (1729-1783) composed more than a hundred works for his royal pupil, the musically talented Infante Don Gabriel de Borbón, among them the six double concertos presented on this CD. The paired organs used for this recording were a long time in the finding. Paired organs are rare but they do exist in Italy, Spain and Portugal. However it wasn’t until the Koopmans discovered a pair in the Basilica della Misericordia in San Elpidio a Mare, Italy, that this recording became feasible. What makes these organs, built in 1757 and 1770, particularly suitable is that they are physically close enough to one another to speak together with great clarity. It’s a good recording on a good label, but I don’t believe I’ve ever before heard quite this level of musical and ambient detail from this CD. If I sit back and close my eyes I find myself in the basilica listening to these lovely 18^th century instruments. The reed stops on these beauties simply have to be heard to be believed. I’ve heard a lot of organs in Europe and America and a lot of organ recordings, but I have never heard reed pipes that sound like this.

Ambient information can run a whole gamut, from the merest hint of venue to a totally convincing presentation of it, as in this case. This is precisely the sort of information that is difficult to retrieve, requiring a very low noise floor and high resolution. (A proper room wouldn’t hurt either.) The retrieval of information below one bit is conceptually challenging, like the noise (dither) that makes it possible. It would be interesting to hear this DDD recording at its original bit depth and sample rate to compare with the Redbook version. Ambient retrieval, vivid partials, convincing dynamic relationships and lucid presentation of the tracker mechanics and valve chuff are all indicators of the excellence of the Mk II.

Franz Schubert: Sonata in B-flat, D.960, Moments Musicaux, Hungarian Melody; Edward Rosser, piano (Connoisseur Society CD4275) Schubert died at the age of 32. As with Mozart, it is an incalculable loss. His late piano sonatas, particularly the posthumous D.960, include some of the profoundest music ever written. I began a correspondence with the pianist, Edward Rosser, almost a decade ago. Already an accomplished concert pianist, he decided in his twenties to relearn technique. At an age when pianists embark on careers, Rosser started all over again. And it took 10 years. I’ve been privileged to hear Ed Rosser’s CD releases over the years and he has continued to grow as a musician, as I hope I have continued to grow in my appreciation.

Mr Rosser approaches Schubert with a deep humility, an unflagging sense of service to the music, and a reverence for the sheer beauty of the composer’s musical soul. It’s a rather fatuous thing to say, but I do believe this is his best disc yet. (Or perhaps it’s just that D.960 suits me at this time in my life as never before.) Mr Rosser does an absolutely wonderful job, particularly in the Andante sostenuto, a movement in which a performer can easily lose his bearings. Rosser does not, neither succumbing to melancholy nor treading too lightly the odd bitter-sweetness typical of late Schubert. In the Allegro, ma non troppo, Mr Rosser elicits breathtakingly beautiful tone from the pianoforte (a Hamburg Steinway); I swear, I’ve never heard anything quite like it. What the Mk II does for this disc is splendid. You’ve heard me say it more than once before, It’s like there’s a piano in the living room. Well, it is. There’s more body, more beautiful tone color, cleaner and more sustained overtones, harmonically richer bass. In a sense one’s whole body, not just one’s ears, are involved. (The documentary film, Touch the Sound, comes to mind.) The physicality of the hammer striking the strings is uncanny. It’s an easy matter to virtually see Mr Rosser at the keyboard, and you don’t have to be in the catbird seat with your eyes closed in order to do so!

Nojima Plays Liszt: Mephisto Waltz No. 1, La Campanella, Harmonies du soir, Feux Follets, Sonata in B minor; Minoru Nojima, Hamburg Steinway (Reference Recordings HR-25) Thirty-two years ago I was actually standing behind the drum-set when Doug Sax recorded Jim Keltner at the Sheffield Labs studio in Culver City. The direct-to-disc finishing process involved playing the vinyl master once through before quickly driving the disc to a Hollywood mastering/pressing lab. (In case you’d like a copy of this acclaimed disc, there’s one on eBay going for $200.) For the first time in my budding career, I got to hear a master source, through what were some the finest tube electronics and monitors of the day, and within minutes of hearing the original performance; a nonpareil A/B opportunity. The reproduced sound was amazing. I don’t believe I’ve heard the like since—not until this disc arrived, courtesy of Jan Mancuso of Reference Recordings.

The original analog recording was made at the Civic Auditorium, Oxnard in 1986 and the digital transfer was made in 24-bit format at a sampling rate of 176.4kHz—256 times the resolution and four times the sample rate of Redbook. At lot of technological water has passed under the digital bridge since that time. But with the recent appearance of music servers and high-resolution converters, Reference Recordings decided to issue some of their masters in digital format. The remarkable HRX series is pressed on 4.7GB DVD media in Waveform Audio File format. Web advertisements for HRX discs caution in red, boldface type: “This disc will NOT play on CD, DVD or SACD Players! – For Music Servers Only!” Not strictly true: there are a few other devices that will handle HRX media, and the Perfect Wave Transport is among them.

It feels almost superfluous to say this is the most revealing recorded piano sound I’ve ever heard. The difference is not at all like the difference between one Redbook CD and another, or even (in my limited experience) like the difference between Redbook and SACD, rather it’s in a different league. Four qualities stand out, though in themselves I don’t think they fully explain this difference: low-level detail, uncompressed dynamics, extended bass, clarity. Yes, I’ve used each of these terms in describing Redbook CDs—which is why the Sheffield Labs story is apposite. In the old days master tapes were regarded as the cat’s pajamas of music sources, as close to the real thing as recorded sound can get. With HD downloads and media such as HRX, that level of reproduction quality is now available at home.

Liszt’s music might be called an acquired taste. It has been dismissed as program music (a derogatory term) full of bombastic posturing by its harshest (and least comprehending) critics, although more so in the past than these days. The terms I have personally used to describe Liszt’s music—heartfelt, emotionally straightforward, ingenuous—have not always been received with credulity. Although one other characteristic—fiendishly difficult to play—is universally acknowledged. Minoru Nojima’s performances are flawless, lucid, articulate, meticulous; his technique, which is simply breathtaking, goes beyond virtuosity. Virtuosity can mask musical shallowness or become a vehicle of self-aggrandizement, but Maestro Nojima uses virtuosity, every astonishing bit of it, in service to Liszt. This sonata is unique in the literature, a continuous, episodic movement of thirty minutes duration. It has been called the greatest thing Liszt wrote.

It would have pleased me to come up with striking, original adjectives to describe my experience with PS Audio’s latest digital front end. I’ve tried my best but I’m stuck with the same old chestnuts. And, I admit, it’s a little frustrating. I was just glancing through my CD collection and Wanda Landowska’s 1949 recording of Das Wohltemperierte Klavier caught my eye—Bach’s greatest keyboard exponent in a very mediocre monaural recording playing her favorite, and my least favorite, harpsichord, a Pleyel. Now I’d mentioned this CD in another review, how this DAC or that cable brought new life to it and made the instrument less strident to my ears. However, the PS Audio takes the biscuit, as though turbulent, muddy waters have been clarified, reifying the Pleyel till it’s almost physically present in the room; behind limiting screens of age and engineering to be sure, but unmistakably there. It’s uncanny and something very special indeed. 

Specifications:
Perfect Wave DAC $3,995.00
Perfect Wave Transport $3,995.00
Website http://www.psaudio.com