Winning the Loudness Wars
At the 133rd Audio Engineering Society (AES) Convention at San Francisco's Moscone Convention Center, a full house flocked to aptly numbered Room 133 on October 27 to hear Stereophile's John Atkinson and four other major audio professionals deliver a two-hour presentation, Loudness Wars: The Wrong Drug? Sharing the stage were the panel's chair, Thomas Lund of TC Electronic A/S from Risskov, Denmark; Florian Camerer of ORF of Vienna, Austria; fabled recording and mastering engineer Bob Katz of Digital Domain in Orlando, Florida; and the equally fabled George Massenburg, McGill University, Montreal, Quebec, Canada, who engineered, among other things, that audiophile favorite, Jennifer Warnes' "Bird on a Wire."
Lund, whose company's LM6 loudness monitoring meter was put to use in slides projected throughout the panel, began with a long demonstration whose message was, in so many words: Don't use the peak level meter to adjust volume. Instead, use the Peak-to-Loudness Ratio (PLR), which measures the peak level of a track ("true-peak") relative to normalization. This preserves micro-dynamics, in which the heart of music lies."
Lund explained that the ear/brain judges loudness by the music's average level. Dynamic Range is not the same as Loudness. The louder the music, and the higher the average level of loudness, the lower the dynamic range, and the more lifeless and squashed a recording becomes. Maximizing loudness actually minimizes transient peaks, to the detriment of the music.
"Don't squash the peaks to control loudness!" Lund proclaimed. As proof, he played four modern tracks that sounded pretty hideous. In fact, one of the high points of his presentation came when he projected a chart that compared the PLR of five well-mastered tracks (Paul Simon's "Homeless," Donald Fagen's "I.G.Y.," Warnes' "Bird on a Wire," Lou Reed's "Dirty Blvd.", and Michael Jackson's 1982 "Beat It") with five others that were pretty awful (Take That's "Shine," Michael Jackson's 2009 "Beat It" remastered edition, AC/DC's "Rock'n Roll Train," Kelly Clarkson's "Since U Been Gone," and Metallica's if only it were true "The Day That Never Comes"). Another chart showed the abysmal decline of the Peak-to-Loudness ratio of the 10,000 most popular music tracks in the US, UK, and Germany from a relatively Golden Age of Recording, which reached a PLR apex of 16dB in 19821984, to the nadir of 20062008, where PLR declined to a little over 9dB.
Engineers to the Fore
Floridian Bob Katz, introduced as "a brother in arms for many years," did his tie-die San Francisco thing by proclaiming that he votes to preserve free-range organic transients. He was joshing, but he wasn't far off. By the time you read this, Californians tried but failed to buck a huge PR campaign spearheaded by Monsanto and Dow Chemical to prevent labeling of foods containing GMOs.
Katz, who has just had his Mastered for iTunes book published, declared, "Per track normalization in iTunes is an insidious compressor. It makes soft songs sound too loud." Far more preferable is album normalization. To make his point, Katz projected a chart that measured the per track normalization of four tracksloud and soft songs by the Beatles and Frank Sinatrato demonstrate that per-track normalization makes soft songs sound way too loud, and diminishes swing in loud numbers. Album normalization enables everything to sound and feel right.
"Should it be the mastering engineer's job to fix the damage that that mixing engineers have done?" Katz asked. "Why should I have to restore the peaks the mixing engineers have squashed?"
George Massenburg, introduced as the man responsible for some of the finest rock mastering Lund has ever heard, made the case for "aggressive use of dynamics." He also played several fine-sounding tracks, including a recent recording by Aimee Mann, on which compression has been selectively employed on individual instruments to positive effect.
Florian Camerer shared good news. As the chair of a panel that sets European broadcast standards, he announced that European broadcasters are switching from Peak Track Normalization to Loudness Normalization. (The European broadcast standard currently normalizes PLR at 23 LUFS/LKFS, and may switch to 24 in December.)
"The tide has turned in Europe in terms of broadcasting," he said. TV is already in the bag, and radio is next. He also noted that iTunes, which now normalizes loudness at 16.2 LUFS/LKFS, and uses peak normalization for tracks with high PLR, is doing a pretty good job.
The Listener's Perspective
JA, whose opening slide was used as the heading image for this story, was introduced as the one panelist who would speak from the perspective of listeners, ie, us. Although time limitations forced him to truncate his presentation a little, John was able to play Shelby Lynne's "Just a Little Lovin'" to demonstrate how true music lovers value space between sounds, and a dynamic range that preserves a huge amount of transient information.
For contrast, he then played an extremely noisy track by the Red Hot Chili Peppers in which space was no longer the final frontierit didn't even exist. "As a Chili Peppers fan, I want to enjoy this," he lamented, "but I can't." People knew how to preserve dynamics in classical music well way back in 1932, he noted, yet, in 2009, Don Was so flattened the peaks on Delbert McClinton's musically excellent album, Acquired Taste, that it couldn't be featured it as a "Recording of the Month" in Stereophile.
JA welcomed the idea of Loudness Normalization in broadcasting, as now those who squashed the life out of music to increase un-normalized loudnessthe current situationwould be penalized, as post-normalization, all the listener would hear would be the damage done to the music: "wimpy "loud" music, in Bob Katz's immortal phrase. The incentive for record industry "suits" to hyper-compress songs during mastering would thus be eliminated.
Turning to "Money" from the hybrid SACD issue of Pink Floyd's Dark Side of the Moon, John compared the SACD layer, which preserves the superior original mastering, to the CD layer, which squashes dynamics. Summarizing observations that he first elaborated upon in his 2008 "As We See It" editorial, "CD Quality: Where Did the Music Go?," he concluded, "As an end user, I welcome the trend in the industry toward reliance on the Peak-to-Loudness ratio. Loud music should sound loud, not weak or wimpy."