CD Player/Transport Reviews

The technical specifications for the Compact Disc include a sampling rate of 44.1kHz at a resolution of 16 bits. The Nyquist Theorem states that if you want to sample a sinewave and then reconstruct it, sampling at twice the sinewave's maximum frequency permits perfect reconstruction of that wave (assuming certain other conditions are met). So a sampling rate of 44.1kHz should take you up to an audio frequency of 22kHz, which is higher than most people can hear. However, for this process of recording and reconstruction to work properly, the bandwidth of the analog signal must be restricted to 22kHz.

But how do you get rid of frequencies above 22kHz? Most audio engineers agree that this is a job for digital filters—analog filters aren't steep enough or consistent enough to do the job effectively. There are basically two families of digital filters: extremely steep "brickwall" filters (the most common), and filters with a more gradual rolloff, both implemented as what are called Finite Impulse Response filters. However, both compromise the impulse response: with a single transient as the signal, the output shows ripples before and after the signal itself, very much so with the brickwall filter. (Most of the points I cover here are based on a white paper available on the Ayre Acoustics website.) Ripples that occur before the signal transient may be particularly problematic because, unlike the post-signal ripples, these "pre-echoes" are not like anything encountered in normal music. It's as if a guitar string began to faintly vibrate before it was plucked. A slow-rolloff filter—provided as an option in such CD players as the Onkyo DX-7555, and by Ayre's CX-7e when set to its Listen position (I wrote about both players in the January 2008 Stereophile)—reduces the pre- and post-ringing, but at the cost of greater leakage into the audioband of frequencies above 22kHz.

The latest attempt to deal with this problem is the "apodizing" filter, described in 2004 in a paper by Meridian consultant Peter Craven, published in the Journal of the Audio Engineering Society, and discussed in John Atkinson's review of the Meridian 808.2/808i.2 Signature Reference CD player/preamplifier in the April 2009 Stereophile. The word apodizing is a curious one. Its literal meaning is "removing the foot" (pod = foot, as in podiatry), which doesn't sound pleasant at all. Wikipedia defines it as "changing the shape of a mathematical function, an electrical signal, an optical transmission or a mechanical structure," which gets us closer to the relevant meaning.

In audio, Craven's apodizing filter is a digital filter with a corner frequency at slightly less than half the sampling rate—in the CD's case, at 22.05kHz. The effect of using the apodizing filter is that the pre-ringing is eliminated, the tradeoff being more and longer post-ringing. My understanding of digital filter theory is, at best, rudimentary, so don't ask me to describe exactly how this is accomplished; suffice it to say that the elimination of pre-ringing and the presence of more extensive post-ringing were confirmed by JA's measurements in his review of the Meridian 808i.2. The greater post-ringing is not ideal, but since this mimics the natural resonances of musical instruments, it's thought to be preferable to the "unnatural" pre-ringing.

In the Minimum Phase (MP) revisions of their well-received C-5xe SACD/CD and CX-7e CD players, Ayre Acoustics has incorporated something similar to the Meridian apodizing filter, but with some wrinkles of their own. (The C-5xe^MP was reviewed by Wes Phillips in the July 2009 issue.) Through measurements and extensive listening tests, they found that the best audible results were produced by combining Craven's apodizing filter with a slower rolloff, and tweaking other aspects of the filtering process. The resulting impulse response is claimed to have no pre-ringing, and only one cycle of post-ringing. As with the CX-7e, a toggle switch in the back of the player gives a choice of Measure and Listen filter alternatives; in the case of the CX-7e, this was a choice between a brickwall filter (Measure) and Ayre's slow-rolloff filter (Listen), whereas for the CX-7e^MP and C-5xe^MP, the choice is between the "plain" apodizing filter (Measure) and the Ayre's modified slow-rolloff version of this filter (Listen).

The design of the C-5xe allowed the change in the digital filter to be accomplished in the player's firmware, but for the CX-7e some hardware changes were required to produce the MP version. The DAC has been reconfigured to bypass its internal filter and to instead use Ayre's own external digital filter. The cable connecting the drive to the circuit board was changed from the standard ribbon to twisted pairs, which reduces the stray radiation of high frequencies that can interfere with the analog audio signal. They've also installed new operating-system firmware for the microprocessor that controls the CD-ROM drive used by the CX-7e. This allows direct access to tracks without having to wait for the CD's Table of Contents to be read; also, a Time Remaining mode was added to the Display options.

I own a CX-7e and was eager to have it updated to MP status ASAP. For review purposes, however, I wanted to be able to do A/B comparisons with the MP version, so Ayre sent me a CX-7e^MP player for comparison. Except as noted, I used both players with their Listen/Measure switches in the Listen position (footnote 1).

Sound

The Ayre CX-7e was/is an excellent CD player—you don't get into Class A of Stereophile's "Recommended Components" by being anything less. When it came to selecting a reference CD player for myself, I chose the CX-7e because I was impressed with its sound compared to the tweaked combo of transport, upsampler, and DAC I'd been using, and felt that I'd have to spend a lot more money if I wanted a player that would represent an improvement on the CX-7e's sound. I was happy to use it with other equipment in reviews, content in the knowledge that the CD player was not the weak link in the system.

Still, it didn't take much listening to the CX-7e^MP for me to conclude that it was simply, well . . . better. This impression was confirmed by back-and-forth comparison with the CX-7e.

How was the CX-7e^MP better? To begin with, it wasn't a matter of a different tonal balance: the CX-7e^MP didn't sound brighter or duller, lighter or heavier in the bass, more forward or more laid-back. And yet every familiar CD I played through the CX-7e^MP sounded more like live music and less like a recording. This was the case even with some very early CD releases, such as the very first CD I ever bought: the original cast recording of 42nd Street (RCA RCD1-3891). I love this recording for the music and the sheer theatricality of its performances, but when the chorus is singing and the orchestra is going full tilt, things get rather congested. This congestion was much less with the CX-7e^MP—individual voices in the chorus became more distinct, and the sound of the chorus tap-dancing was more like the real thing. Orchestral recordings featuring massed strings seemed to particularly benefit from playback though the CX-7e^MP, getting closer to the sound of the real instruments rather than sounding like synthesized versions thereof. Very impressive.

Given the apodizing filter's elimination of pre-ringing, one might expect that instrumental sounds with sharp transients would be major beneficiaries of this design—and that was indeed the case. A notable example was the cymbal at 0:55 of Ana Caram's "Viola Fora de Moda," track 3 of the Chesky Records Jazz Sampler & Audiophile Test Compact Disc, Vol.1 (Chesky JD37). This is one of my CD test tracks, and I must have heard it literally hundreds of times. With the CD played back through the CX-7e, the cymbal, which is played pp with a soft mallet, sounds well recorded, with a nice shimmer and a sense of space around it. However, playback through the CX-7e^MP resulted in an even more natural onset and decay of the struck cymbal transient, and the "air" around the instrument was better defined. The placement of voices and instruments in the space of the recording venue was generally more precise, an effect that was evident on all recordings.

When I first listened to the CX-7e^MP, my initial impression was that it sounded a bit louder than the CX-7e. The output-level specifications of the two players are the same, but I wondered if there was a difference in the output levels due just to sample variation. I measured the output of the two players (playing the 1000Hz test tone track on the first Stereophile Test CD (Stereophile STPH002-2), measuring the AC voltage at the speaker terminals with a voltmeter), and there was absolutely no difference. So why did the CX-7e^MP give the impression of sounding louder? My guess is that this was due to the more sudden onset of transients without the pre-ringing.

In reviewing the CX-7e, I compared its Measure and Listen options, and much preferred Listen. I made the same comparison with the CX-7e^MP, and although I once again preferred Listen, this time the preference was not as great. The main difference in sound was that the Measure setting sounded brighter, more forward—but by no means unpleasant. Overall, I preferred the more "refined" sound of the Listen setting, but I can imagine that in some systems Measure might be preferable. At least you have the option.

Should You Upgrade?

Because upgrading the CX-7e to MP status requires hardware changes, the unit must be returned to the factory. The CX-5e MP update—which, in addition to the firmware change, includes a full checkout and adjustment of laser alignment, if necessary—costs $200, whereas the cost of the CX-7e MP update, which includes all that plus the hardware changes, varies from $250 to $900 (the latter only for some very early units). I'd say the money is extremely well spent, and I urge every CX-7 and CX-7e owner to get in touch with his or her local Ayre dealer to arrange the update. The price of a new CX-7e with the MP filters remains at $3500, and the addition of the MP filter options make it an even better buy.—Robert Deutsch