New Media Metrics Conclusion
I empathize with those readers who have contacted Stereophile asking for some summary of the graphical evidence I presented in "New Media Metrics." I had planned to include a more rounded conclusion to the piece, but, with my word budget almost spent and a guilty conscience at asking John Atkinson to find space for more than 60 graphs, I put a perhaps premature sock in it.
So what messages did all those graphed wiggles convey? Well, let's first admit that, substantial though the exercise was, it can make no claim to represent a statistically significant sample of commercial SACD and DVD-Audio discs. There are doubtless some even more impressive spectral contents to be found, along with some spectacular duffers as yet unmasked. I can only speak of the selection of discs I measured—or, even more specifically, of the tracks on those discs that I selected for the analysis.
On the positive side, I was gratified to find more than the occasional example of spectral content reaching out to beyond 40kHz. Many studio microphones have a high-frequency response that falls away beyond (or even before!) 20kHz, and this characteristic can be seen in some of the results (for example, the Friendship SACD). But some engineers are clearly making the effort to use microphones with a more extended HF capability. Whether this matters is one of the continuing uncertainties surrounding high-rate digital audio, but when the extra bandwidth capability is available, it seems at least miserly not to exploit it.
On the subject of uncertainties, SACD's high level of ultrasonic noise—amply demonstrated in both the spectra and the rate-of-change graphs—continues to be a conundrum. Does it have a subjective effect even if the downstream equipment is tolerant of it? I don't think we know yet. What can be said with confidence is that, on some recordings (eg, the Steve Davis disc), there is clear evidence of ultrasonic music content being swallowed up in SACD's shaped quantization noise, which might just be an argument for saying that the medium's effective bandwidth capability is insufficient.
Something I discovered the hard way in compiling this report is that too many DVD-A and SACD releases tell you not nearly enough about the origin of their content. CD's three-letter SPARS code—which indicates which domain, analog or digital, the recording, mixing, mastering, were done in (AAD, ADD, DAD, DDD)—tells you only so much, granted, but it's more than some hi-rez releases do. Hell, one of the discs I chose didn't even include track timings, either on the tray card or in the booklet, which by any account is plain inept. You can appreciate, though, why record companies would sometimes prefer to be coy about such things. Discovering that the Alison Krauss SACD was derived from a 48kHz PCM master, for instance, makes you scratch your head in wonder. And I doubt that Reprise would be keen to concede that the Frank Sinatra track had been ineptly upsampled at some point—although that is what its spectrum clearly indicates.
If I ruled the world, every SACD and DVD-Audio disc reviewed in the audiophile press would have its spectral content laid bare so that issues like this become inescapable for the culprits. Such a practice might, if nothing else, encourage the record industry as a whole to be a) more upfront, and b) more careful about what it lays out for our delectation. And our bucks.—Keith Howard