The Fifth Element

I have no inside knowledge, but I wouldn't be surprised to learn that the compression used on Everything Must Go was digital, whereas the compression on The Nightfly had to have been of the more benign analog kind, because digital compressors were not yet available in 1982. Indeed, The Nightfly was one of the first digitally recorded rock albums. (The first one was Ry Cooder's Bop Till You Drop, in 1979.) Digital recording was comparatively new, and there were as yet no digital effects boxes (footnote 5). (Trivia: The Nightfly, perhaps for its feeling tone of bittersweet but optimistic yearning, made it onto the "Top 10 Albums List" of Vatican City's newspaper, L'Osservatore Romano.)

In theory, using extreme dynamic-range compression might make sense, if your goal is to have your record "sound loud." In comparisons, untrained listeners usually respond more favorably to the louder sound—up to a limit. (This preference is why, in critically assessing equipment, it is important to match volume levels.) But in practice, it just doesn't work. The fundamental reason that overcompressing dynamic range to make music sound louder is self-defeating is that you, and not the record producer, are in control of your own stereo's volume control. If a song has too much dynamic compression, it will sound unnatural and be fatiguing to listen to, which means you're likely to turn the volume down to a more reasonable level. Once you have, the lack of difference between loud and soft that results from overuse of dynamic compression will make the music sound muffled. To quote Matt Mayfield's concise diagnosis: When there's no "quiet," there can be no "loud."

Mayfield has produced a two-minute YouTube video that is a concise and comprehensible demonstration of the self-defeating nature of maximizing volume through overuse of dynamic compression. He applies extreme compression to a song that was released without it: "Figure of Eight," from Paul McCartney's Flowers in the Dirt (1989). You can clearly hear the loss of impact even through loudspeakers as humble as those that come free inside an iMac. The video, "Loudness War Explained," has been viewed more than a million times on YouTube.com, and is so persuasive that mastering guru Bob Ludwig uses it when he lectures to professional groups. Masterer Bob Katz, who worked on many of Chesky Records' well-regarded early releases, generously hosts a higher-resolution version of the video here; I urge you to check it out.

So: overuse of dynamic compression is a bad thing, and there should be much less of it (footnote 6). But what can be done? Well, first, there's much to be said for education and the bully pulpit, not only in changing public opinion, but also in influencing the decision makers. Even more important, practical help is on the way.

Several years ago, the International Telecommunication Union, largely in response to consumer complaints about television volume jumping between program content and commercial messages, formed a study group to examine metering and loudness control. Last August, their report resulted in Recommendation 128 (R128) from the European Broadcast Union: that metering should change from Peak metering to Loudness metering. In other words, Instead of focusing on how nearly the program material's peak level approaches 0dBFS (Full Scale), metering will, in a sense, "float," with the program material's dynamic range itself indicating, as a long-term average, what the "center of gravity" of the program's loudness is. (If you want to read the ITU and EBU documents, they're linked to at the end of Wikipedia's article on "Loudness War.")

The two following graphs illustrate the practical difference: Using Peak metering, different program material ends up with widely varying degrees of loudness, from classical to hyper rock. Referencing long-term average Loudness, however, and normalizing the diverse program material to have the same Loudness "center of gravity" (differing now only in width of dynamic range), the huge difference in headroom between classical music and hyper rock no longer skews the results.

TC Electronic (Denmark) has already released the LM2, a stereo loudness and true-peak level meter that complies with EBU R128 (see www.tcelectronic.com/lm2-features.asp).

I heard Bob Ludwig's latest presentation on developments in metering and loudness just days after the US Congress had passed a law requiring that television commercials not sound louder than program content (footnote 7). Ludwig's presentation was fascinating, as is usual for him, and one of its most fascinating aspects was his projecting onto a screen the meter bars and "radar screen" sweep of TCE's LM2 loudness meter for various kinds of music content as it was played.

Ludwig played some 10 musical examples. One was the slow movement from a Beethoven piano sonata. Another was "I.G.Y.," from Donald Fagen's The Nightfly. Both sounded great. All the other musical examples were so compressed that they sounded like crap.

Ludwig's hope, which he shares with hundreds if not thousands of conscientious audio and recording professionals worldwide, is that, in the not-too-distant future, when people buy music downloads, each download will come with a new piece of metadata derived according to R128. When the music is ingested by a program such as iTunes with Loudness Normalization engaged, for whatever setting the listener has set his or her volume knob to, the server or iPod will know that, say, the overall level of the slow movement of a Beethoven piano sonata must be raised 8dB, while the overall level of the Metallica track must be lowered by 14dB, so that all tracks on the playlist have the same subjective loudness. Once this becomes "the new reality," there will be much less of an incentive for a producer to tell an engineer to compress the crap out of a song, because that song will be analyzed according to R128, the metadata won't lie, and the server will automatically reduce the music's level to correspond to the listener's preferred volume setting. Gotcha!

I heard Ludwig's presentation at the annual Parsons Expo 2010 in Dedham, just outside Boston. Details of Expo 2010 can be found here.

Thanks to Matt Mayfield for his screen shot, to Thomas Lund of TC Electronic for his graphs and screenshot, and to Bob Ludwig for his fighting the good fight and for his technical comments on an early draft of this.

A Reminder
Due to the recent remodeling of Stereophile's website (the layout is cleaner, and more invitingly presents the content), the winning lists from my "Mystic Chords of Memory" write-in contest went up a bit late, then became inaccessible for a few weeks. If you went looking but found nothing, please check out the winners here. Lots of rewarding music there.

Questions or comments.

---

Footnote 5: Don't necessarily blame the mastering engineer for overuse of dynamic compression; the musicians or the producer might have been responsible.—John Marks

Footnote 6: I grow weary of explaining to neophyte and veteran audiophiles alike that a recording that sounds "quiet" is not suffering from a "lack of dynamics," but on the contrary has a wider-than-usual dynamic range. Having to accommodate those large transient peaks within the 16-bit window of the CD's encoding pushes down the average level of the recording, hence the loudness.—John Atkinson

Footnote 7: The law is called the Commercial Advertisement Loudness Mitigation Act, or CALM Act.—John Marks