Nick Southall has a fascinating and controversial article over at Stylus Magazine about another incredibly annoying trend in pop recording (and, increasingly, in nonpop recording) -- the loudness arms race, where records are mastered with increasingly absurd levels of compression in order to make them sound "hotter" than everything else. This process also flattens out all the dynamic peaks and valleys, making the entire record unremittingly loud and, often, clipped and muddy as well. Many recording engineers get really steamed about this -- as they should, because too much compression (especially during the mastering process) totally ruins all of their meticulous work in the studio -- but they are under constant pressure from artists, producers, and labels to "make it hotter."
Record companies these days (and I don’t just mean nasty behemoths like EMI or Sony—your favourite indie are probably just as bad) are eager to make CDs as loud as possible because they think, with some justification, that this is what people want. In order to get CDs to be consistently loud, they get compressed—essentially this means that the quieter moments are made louder in relation to the, um, louder moments, to make the entire CD a consistent, and high, level of volume. During the compression process, the tops of signals can be cut off, or "clipped." Compress a record too much, and it sounds bad. Make it “clip” even slightly, and it sounds worse.
There are two ways to measure "loudness"—peak levels and average levels. The former refers to the loudest part of a piece of music or sound; a crescendo or climax. The difference between the highest and lowest points makes for the average level. Sadly, the science of psychoacoustics suggests our ears generally respond to the average level rather than the peak level of volume—hence we would perceive a consistently loud piece of rock music as being "louder" than a piece of classical that reaches the same or even a higher volume level during a crescendo, simply because the rock song is "loud" all the way through. "Loud" records grab our attention (obviously—being louder they are harder to ignore on first impression) and in order to grab attention quicker and more effectively in a crowded marketplace, record companies and artists have been striving to make their records as loud as possible from the second the first note is played, whatever the cost.
Levels have crept up over the last decade though, and alarmingly so. Nevermind is 6-8dB quieter than, say, Hopes & Fears by Keane—to contextualise this, those 6-8dB will make Nevermind sound approximately half as loud. On most modern CDs the music is squashed into the top 5 dB of a medium that has over 90 dB of range. It’s like the oft-quoted fact that humans use only 10% of their brain [N.B. Not actually true - DJA]—imagine what we could do if we realised potential. Think of the classic, exciting Pixies formula again—it doesn't exist anymore, because those dynamic leaps have been ironed out. Keane should NOT be twice as loud as Nirvana.
Music with an incredibly loud signal is referred to in the industry as "hot." One way to make music "hot" is by compressing it—essentially this means lowering the peaks so they're almost level with the troughs, and then increasing all of the signal to make it as loud as you can before it starts "clipping." Only a lot of people seemingly don't know when to stop. Compression can be added at almost any stage of the recording process, in large or small doses. Small increments added at both recording, mixing and mastering are more effective in preserving sound quality than huge leaps taken at the final stage, for instance.
One result of this is that modern CDs have much more consistent volume levels than ever before. But when is it desirable for music to be at a consistent volume? When it's not being actively listened to; i.e. when it's intended as background music. Sudden (or even gradual) dynamic changes in ambient volume disturb people from what they are otherwise doing (shopping, eating, working) by making them pay attention to the fluctuating sound rather than the task in hand—I only notice the air conditioning at work when it switches off ten minutes before I go home every day, for instance—for the previous eight hours, my brain tunes the hum out so I can concentrate. So it is with music too—it may grab your attention more effectively at the start, but it's ultimately easier to ignore too. All music becomes background music if it's at one flat level, no matter how loud.
If there’s a jump-the-shark moment as far as CD mastering goes then it’s probably Oasis. In 1987 Appetite for Destruction averaged about -15dB RMS volume, and was considered loud. By 1994 the average loudness in RMS power for a rock record was -12dB. (What’s the Story) Morning Glory in 1995 hit a phenomenal -8dB on many tracks. The 1997 remaster of Raw Power reaches an extraordinary -4dB, making it supposedly the loudest rock record ever. In 2005 the average RMS volume is -9dB. Audiophiles and people who work in audio engineering largely agree that this is too loud, but in the face of massive commercial impetus their say is often ignored. Arguably (What’s the Story) Morning Glory became so successful in the UK precisely because it was so loud; its excessive volume and lack of dynamics meant it worked incredibly well in noisy environments like cars and crowded pubs, meaning it very easily became an ubiquitous and noticeable record in cultural terms.
On At War with the Mystics, for instance, there is so much clipping during the crescendos (which aren't real crescendos anyway, because they're the same volume) that it almost seems as if it's being used deliberately as another instrument in the mix. It’s this flatness, this clipping, this unwavering attack, that wears and tires and means you won’t listen to your favourite records, if they’re from the last few years, as often as you might want to, because they are intrinsically unmusical and unpleasant. Hence, perhaps, the perpetual merry-go-round of seeking the newest flavour-of-the-month; over-compressed music sounds great for a couple of listens, but there is little desire to replay the music because your brain recognises that there is something fundamentally unmusical about the sound.