Dynamics & Dynamic Range

One of the delights of being published by a multinational conglomerate that grows through acquisition, as Emap Petersen does, is that Stereophile finds itself in interesting company. Like La Nouvelle Revue du Son in France, for example, edited by the legendary Jean Hiraga, who turned me on to the sonic importance of wires and passive components almost 25 years ago. And Mojo, an English music magazine tightly targeted on baby boomers like me, who bought their first stereo systems in the '60s to better appreciate the progressive rock we lived and loved by. (I wonder if turn-of-the-millennium college students gather 'round a new G4 Mac to get off on MP3s the way, 30 years ago, we gathered 'round our precious vinyl.)

As well as a great piece on Miles Davis' Kind of Blue sessions, September's Mojo had an in-depth piece on Carlos Santana. (See what I mean about targeting?) This culminated in praise for the guitarist's new CD, Supernatural (Arista 19080-2), which features all manner of famous friends, from Eric Clapton (for people my age) to Lauryn Hill and Dave Matthews (for people like my teenage daughter). So, being a well-trained consumer, I called in at Borders on my way home from the office and bought a copy.

I knew I shouldn't have bothered, that instead I should have just played my Japanese pressing of Caravanserai one more time. Musically, Supernatural isn't bad, but the sound is a different story. Most of the cuts are compressed to hell. They sound loud when played at low levels, but when you turn up the volume the relentlessness of the sound, the total lack of dynamic light'n'shade, have you turning it down again.

Recordings like Supernatural are anti-hi-fi. There is nothing more to be gained from playing them back on anything with greater pretensions than a boombox. They're also anti-musical, in my opinion. If everything is at the same level, then how can there be any musical interest? If listeners are thrilled by the occasional loud climax, that doesn't mean that sustained loudness is continually thrilling—sorry, Red Hot Chili Peppers fans!

I checked out Supernatural's dynamics with a Dorrough AES/EBU meter, which shows peak and average levels. Like those of almost all CDs these days, Supernatural's peak levels lit the 0dBFS LED on the meter (which actually indicates 3LSBs below maximum level). All mastering engineers do this to squeeze the maximum resolution out of the CD medium, and it's what I do with Stereophile's CDs. But the Santana CD's average levels (by which the ear judges loudness) ranged from -5dB to -8dB (footnote 1), revealing that someone in the recording/mixing/mastering chain had passed it through an aggressive compressor/limiter to make it sound "loud."

By contrast, I recently received an e-mail of complaint from a reader who had bought Stereophile's Rendezvous CD. "Your CD has no dynamic range," he wrote. "It sounds quiet." I sighed when I read this. One of the great audio confusions is that of "loudness" with "dynamic range." The reason you have to turn up your volume to play Rendezvous is because it has enormous dynamic range (footnote 2). Its average levels are around 20dB below its peaks. But this low average level means that it sounds "quiet" compared, for example, with the Santana CD, which has very little dynamic range.

When I bought the Santana disc, I also picked up a copy of one of this issue's "Recordings of the Month," the Beatles' Yellow Submarine Songtrack (Apple/Capitol CDP 5 21481 2). John Swenson's review (p.171) is right on the money: The engineers at Abbey Road Studios have indeed done a great job of remastering. Again, the CD peaks at 0dBFS, but while the average level does occasionally reach -6dBFS, it is often around -12dBFS or even lower. Compared with Supernatural, this 30-year-old recording does have a semblance of dynamic range.

Just for grins, I measured the peak and average levels of Yehudi Menuhin's seminal 1932 performance of the Elgar Violin Concerto, recorded, again, at Abbey Road (EMI Classics CDS 7 54564 2). Obviously, by its very nature, classical music has much more light and shade than rock. But in the orchestral climaxes I recorded typical average levels of around -11dBFS, with peaks of 0dBFS. It's ironic that, at the turn of the millennium, a 68-year-old transcription from 78s can have better fidelity, in terms of dynamic range, than a typical 1999 rock recording.

Before you all write in to dis me as an old fogey who hates rock music, please note that I grew up on rock music, that I played and recorded rock music professionally, and that I think the best that rock has to offer is music for all time. It's just that I also love the sound of live musical instruments, and I'm getting tired of the overcompression that is being foisted on us. I may have a sweet tooth, but that doesn't mean I want a pound of sugar in every cup of coffee!

These thoughts were triggered by an excellent lecture by Bob Katz, the recording engineer responsible for some of the excellent Chesky CDs and DADs, given at the Audio Engineering Society Convention in New York at the end of September. Space is too short here to go into detail about Bob's presentation, which tied together the concepts of peak and average metering and calibrated monitoring levels in order for mastering engineers to produce CDs with volume levels that are heard as consistent from disc to disc. But he, too, is distressed by the phenomenon of ever-shrinking dynamic range on recordings. You should check out the essays on his website—they pack a lifetime of audio wisdom into a minimum number of words. See also Peter van Willenswaard's "Industry Update" in this issue (pp.27-31), in which he replicates some of Bob's tests on high-sample-rate digital audio.

Triggered by Michael Fremer's Follow-Up (p.49) to his KR VT8000 MK monoblock amplifier review in November, to which Dr. Riccardo Kron has penned two impassioned "Manufacturers' Comments" letters (pp.189-193), I had intended to devote this "As We See It" to the topic of measurements. However, all that it is necessary to say is that, while measurements can't replace or even describe the listening experience, they do shine the light on areas that need clarification or investigation, as in the case of CD dynamic range.

Footnote 1: The average level of a sinewave is 3dB below its peak level. When you listen to recorded music with a peak/average ratio of less than 3dB, you're listening to harmonic distortion. Some of the cuts on the Chili Peppers' Californication album feature a peak/mean ratio of 4dB, the lowest I have measured so far.

Footnote 2: Rendezvous, featuring bassist Jerome Harris' Jazz Quintet, was shortlisted in four categories for the 42nd Grammy Awards: Best-Engineered Non-Classical Recording, Best Jazz Performance, Best Jazz Composition ("Only Then"), and Best Jazz Arrangement ("The Mooche"). We will find out on January 4, 2000 if it made it through the first round of voting to be nominated in any of these categories.