The Great XRCD Shootout of ’04

Russell Lichter

Thanks to Kevin Berg of Japan Victor Corporation of America, I was finally able to do something I’ve wanted to do for a long time: a comparison between an XRCD disc and a non-XRCD disc of the same recording: Bartok’s Concerto for Orchestra, with the Chicago Symphony, conducted by Fritz Reiner [JVC JMCXR-007 and RCA 0926-61504-2]. And I want to let you know what I heard.

I’ve commented on the music in a previous review. Bartok can be a bit ponderous, his melodic inventiveness ranges from the beautiful to the strange, but this music is among his very greatest, a work of genius, performed by a great symphony orchestra conducted by a man who was a life-long friend and proponent of the composer. It is widely viewed as the classic performance of the Concerto for Orchestra. However, to return to my purpose:

JVC’s XRCD reissues of classic analog recordings have a reputation for exceptional sound. Some audiophiles regard them as the best digital sound money can buy. I too have found XRCDs to be consistently outstanding and I’ve bought a number of their jazz CDs over the years. Relatively few since I’m only a very occasional jazz listener. But I’ve never before been able to make a direct comparison with a “standard” reissue, and I was especially excited about this opportunity because this is not only classical music, it is classical music I particularly like.

You will know that JVC begins with analog master tapes that are remastered, digitized and manufactured under conditions that might be described as fanatical. Concerto for Orchestra is one of RCA’s earliest experiments with multi-track recording. It was originally recorded in 1954 using Neuman U-47 cardioid and M-49/50 omnidirectional microphones, summed through passive electronics, using no equalization, feeding an Ampex 300-3 half-inch tape recorder. All audio equipment, of course, used vacuum tubes in those days.

For this XRCD reissue new master tapes were made at BMG/RCA Studio in New York City from the originals, which had deteriorated. There were problems with oxide flaking, defective rewinds and material fatigue. (Think of the master tapes of great, classic performances sitting in vaults, deteriorating, developing print-through, and the apparent indifference of their legal owners who seem to have no interest in saving them to digital format.) JVC engineers used a fully restored Ampex 300-3 for the reconstruction and playback. Employing the XRCD2 process and proprietary electronics, they converted the analog master tape to 20-bit word-length binary data, storing this on a magneto-optical disc. This disc went to JVC’s Yokohama manufacturing plant, where word length was converted to 16-bits immediately before being encoded and sent to the burning laser. At all stages extreme care is taken to eliminate jitter. JVC’s choice of an aluminum substrate over gold is debatable, but the role of jitter in digital reproduction is not. The album notes make this remarkable statement, “The manufacturing chain [used in producing CDs] is not standardized, and while digitally correct [italics mine], does not always reproduce the highest quality audio possible.” The implication is quite clear: the digital information on the RCA reissue is measurably identical to the digital information on the XRCD reissue.

But they do not sound the same. I can think of four, possibly five, factors that might account for this (bearing in mind that my digital expertise is such that I continue to get bogged down in the numbering systems in Chapter One of Ken Pohlman’s Principles of Digital Audio): a difference in the intensity and spectrum of jitter, a difference in noise level, a difference in sampling procedures used in conversion, and a difference in the dithering algorithm. JVC use a proprietary Extended Pit Cutting Technology to optimizes the linear velocity of the glass master, resulting in precise pit lengths. They also stamp directly from the glass master (limiting the run of XRCDs). These relate to the possible fifth factor: a physically more accurate disc ought to result in fewer read errors on playback, though error correction takes care of the errors that typically occur. Serious consideration of such matters is best left to those fluent in digitalia.

The upshot is that the XRCD in playback reveals information lacking in the non-XRCD. (And bear in mind that I am using one of the great CD transports, an Accuphase DP-90, and a Bel Canto DAC2, reputed to be immune to jitter above 2Hz.) This subtle information presents a palpably greater sense of presence, which is to say, the XRCD sounds more like live music. People passing through the living room while this disc is playing are immediately taken with the sound, in a way that doesn’t happen often.