What's Going On Up There? Follow-up part 3
Figs.9 and 10 show spectrograms for the introduction of the Gershwin Piano Concerto in F (Classic DAD 1018) and for an extract from the Scherzo of Dvorák's "New World" Symphony, from the unofficial Japanese DVD-A sampler. Both were mastered at 96kHz; though the former is from an analog tape and the latter from a digital tape, they don't differ appreciably in their ultrasonic content, which extends to about 30kHz, with brass and cymbals the primary source of this content. Chabrier's España, a 96kHz Erato recording on the official Warner DVD-A sampler (fig.11, to the left of the vertical white time line), was very similar, revealing that it was derived from a true 96kHz master.
Fig.9 Spectrogram of Gershwin Piano Concerto in F, from All the Works for Orchestra & for Piano & Orchestra, Leonard Slatkin and the St. Louis Symphony Orchestra, Jeffrey Siegel (piano), (Classic DAD 1018). Original DVD-V data recorded at 24-bit/192kHz.
Fig.10 Spectrogram of Scherzo, Dvorák's Symphony 9, from an unofficial Japanese DVD-Audio sampler. Original DVD-A data recorded at 24-bit/96kHz.
But I was puzzled by Johann Strauss, Jr.'s Emperor Waltz, from Nikolaus Harnoncourt and the Berlin Symphony on DG, which follows the Chabrier (fig.11, right-hand side). The recorded level is low much of the time, and the orchestration doesn't lend itself to the production of much ultrasonic content. But, as you can see, the spectrogram features both horizontal HF lines and some repetitive UHF "ghosts." Fig.12 shows the spectrum of the Strauss during one of these moments. In the high-frequency part of the spectrum many discrete tones are apparent, the lowest in frequency around 14.5kHz, with groups of tones spaced about 4.5kHz apart. I have no idea what this behavior is due to—intermodulation? aliasing?—but it hardly seems a worthwhile use of the extra dataspace offered by DVD-A just to record spurious HF tones (footnote 2).
Fig.11 Spectrogram of Chabrier's España, with the L'Orchestre Lamoureux, and Johann Strauss, Jr.'s Emperor Waltz, Nikolaus Harnoncourt and the Berlin SO, both from the Warner DVD-Audio sampler. Original DVD-A data recorded at 24-bit/96kHz.
Fig.12 FFT-derived spectrum of excerpt from fig.11 data (linear frequency scale, 0Hz-48kHz, 6dB/vertical div., left channel cyan, right channel magenta).
I started off this piece by intending to write a Follow-Up on the Technics DVD-A10, which strikes me as offering a good value, with its ability to play back CDs, DVD-Vs, and DVD-As [especially at its spring 2001 price of $400]. And, as I said, the real proof of the concept will come when we examine its surround-sound performance. But I ended up doing a Follow-Up to my October 2000 article on the ultrasonic content of recorded music. [An analysis of the ultrasonic content of SACD recordings can also be found in my measurements sidebar accompanying Jonathan Scull's review of the Accuphase DP-100/DC-101 SACD player.]
So what conclusions can we draw?
First, true 96kHz-sampled commercial recordings are few and far between, particularly when it comes to nonclassical music. While this magazine has found that upsampling low-bandwidth digital recordings to 96kHz and 192kHz can make them sound better, it is a waste of the space on a DVD-A disc to use that space to store upsampled data. (It should be noted that the respected mastering engineer, Bob Katz, disagrees.) It also doesn't seem honest to label a commercial recording that has been processed in such a manner as "96kHz," even if that is the sample rate on the disc. Of the seven nonclassical 96kHz tracks on the Warner sampler, fully three appeared to have been originally recorded digitally using a 48kHz sample rate.
Second, as with the Strauss sample, there seems little point in using a higher sample rate if all it does is unmask high-frequency problems with the master that might well have been better filtered out of existence by a lower sample rate.
Finally, it gives me a new respect for the venerable analog medium: Many of the recordings I looked at that turned out to have significant ultrasonic content were originally recorded on analog tape.—John Atkinson
Footnote 2: Subsequent to the publication of the review, I began to suspect that these spuriae are due to watermarking, added after the initial mastering, which, I had been assured by the engineer involved, had featured a clean ultrasonic spectrum.