The Loudness War

The loudness war is a weird k-hole I recently fell into, trying to find out more information about a Stooges remaster.  Let me explain.

Okay, have you ever owned some album, and listened to a song a million times, and then heard the same song on an FM radio and it somehow sounded different?  It was probably because the station used dynamic range compression in their outbound rig.  Here’s my best attempt at a no-math explanation of compression.  Let’s say a sound wave is a bunch of waves, ranging in strength from 0 to 11.  (I was going to say 10, but, Spinal Tap.)  So a song, in some greatly simplified form, would be something like this:  “0 8 11 8 10 2 3 7 7 8 11 0 10 7 2 4”.  If you wanted to make the song sound way louder, the obvious way to do it would be to change that to “0 16 22 16 20…” and so on, but 11 is the most you can get.  But what you can do is boost the lower numbers, and keep the higher numbers the same, and the song will “seem” louder.  So, you’d run it through some magic digital box and it would change to something like “0 8.5 11 8.5 10 7 7.2 8 8 8.5 11 0 10 8.5 7” or whatever.  It basically smooshes the lower end frequencies, and makes the song seem louder at the same volume, although this sacrifices some of the sound quality, which isn’t as big of a concern when you’re just schlepping pop music across the airwaves and you want your station to get the most attention when someone is flipping through channels.

The loudness war started back in the days of jukeboxes, because you the consumer can’t change the volume on a jukebox, and everyone wanted their 45 record to sound the loudest.  On a digital CD, that magic number 11 I mentioned above is called “full scale”, or the point where signal has reached as much as it can go.  A measurement called dBFS, or decibels relative to full scale, is used to measure levels, where -6 dBFS is 50% of full scale.  Most albums were mastered with -14dBFS being used as the highest peak level of the album, or what used to be the “red zone” of an analog record.

At some point in the 90s, the thinking changed on this, probably around the time record companies started re-releasing old albums, so if you bought that Iron Maiden album on CD in 1988, you suddenly had to buy the remastered version in 1996.  Yes, they would fiddle with bonus tracks and new artwork a fake gold CD and yes Ray, they included that fourth side of the Live After Death album you bitched about for twenty years, but they also fiddled with the mastering so the “hotter” album would make the old master sound wimpy.  And new albums started getting massively over-compressed in this loudness arms race.

I started googling all of this because of the 1996 Columbia remaster of the Stooges album Raw Power.  When recorded in 1972, Iggy Pop did the initial mastering himself, and through the magic of heroin, decided to put all of the instruments on one side and the vocals on the other, and do all kinds of weird shit with the tone.  The record company refused to release it unless it was remixed, and got David Bowie to spend a single day in a crappy studio, getting the album to sound mostly normal.  For the 1996 re-release, they gave Iggy free reign to go back and remaster the album, and his response, in an attempt to bring back the raw aggression of the original recording, was to completely turn every knob to 11.

I don’t know if this is still the case, but when I bought the 1996 remaster on CD (back when people bought CDs), it had a huge warning label on it, saying the CD did not meet the Phillips Red Book standard and could destroy your equipment.  I thought that was cool, threw it in the player, and set my volume at something marginal, like 5 out of 10.  “Search and Destroy” came on, and it sounded like someone had replaced my speakers with those paper-cone things you got with the stock Delco stereo on the AM radio of a 1981 Chevette, except with pencils jabbed through them.  Within ten seconds, my receiver SHUT OFF with a strange error message on the display, and I had to unplug it from the power, wait the longest 60 seconds of my life before restarting it, almost certain I’d bricked my stereo.  I was only able to listen to the CD by ripping a copy to MP3 first, which I guess just clipped the hell out of it in the computer’s digital-to-analog conversion.  Still a great album, though.

Another Stooges fun fact:  if you really like the album Fun House, you can go over to iTunes and pick up a complete collection of everything they recorded for the album.  It’s a 142-track “album” that contains every take of every song, plus all of the studio dialogue recorded – basically everything that ran through the sound board back in 1970.  At $99.99, it’s definitely in the “do not accidentally click purchase” category for iTunes.  I haven’t bought it – I think if it was $30, I would be tempted, but I know I would only listen to it once or twice.

  • Not to be a total dick about this, but you're slightly incorrect about how this kind of compression works. (Although you're close enough and maybe just didn't want to go into the kind of detail that almost no one would care about.) It's not just the the quiet parts are made louder. The loud parts are also made comparatively less loud, so that the dynamic range is reduced overall. Which is to say that there is less difference between the peaks and valleys of the compressed waveform than the uncompressed version. Then this compressed wave is pushed as far into the "red" as desired. So, such a compressed song doesn't simply "seem" louder, the average level is higher.

  • jkonrath

    Agreed, it's more complicated than my explanation. Probably the best way to look at it is to look at the "loudness war" article on wikipedia, which has some pictures of sound waves that show what I'm failing to explain here.

    The best example they show is the Metallica album _Death Magnetic_, which was mastered "loud" but which was also released in a less-"hot" mastering for the Guitar Hero game. If you look at the two sets of waveforms side-by-side, it shows the phenomenon. (It also saves you from having to actually listen to _Death Magnetic_, which I would definitely try to avoid.)

  • You explained it probably as well as the typical Science Channel show would have, without getting too bogged down in technical details.
    I mean that to be complimentary, even if it doesn't read that way.