73-!L!*!O!*!U!*!D!*!N!*!E!*!S!*!S! !W!*!A!*!R!*!S!

Better articles than mine

The issue of “Loudness Wars” just got tremendously more complicated with an amazingly technical and detailed article in the Sept 2011 issue of Sound On Sound magazine. Among other things it analyzes 4500 pop tracks going all the way back to the 1960s, and goes through a detailed examination of the statistics of Death Magnetic. Unfortunately he doesn’t go so far as to compare the Guitar Hero version to the commercial one, but I think that’s because it’s not really a released version.

I can’t summarize all his findings in this article; in fact I probably need to read it a few times more myself. But I was excited to see something with so much numeric detail published.

(July 26, 2009) I knew about this article some time ago, but I ran across it again while going through my bookmarks. It was on the austin360.com site back in October 2006. (Austin360 is affiliated with the Austin American-Statesman.) “Everything louder than everything else” This article goes into a great amount of depth about this issue. He goes into the technicalities of hypercompression, and why it’s been an escalating problem (in two words: lifestyle and marketing.) He provides a detailed but understandable discussion about why RMS is a good quick measure of how badly a particular piece of music is compressed.

It was written before Metallica’s Death Magnetic came out, but except for the fact he doesn’t have that example to give, everything he writes is still appropriate.

In fact it’s so good I probably ought to just skip trying to write my own take on it…but since I’ve already started it:

OK, back to my original article…

The page title is sort of annoying to try to read, isn’t it? That’s on purpose, we’ll explain why later.

If you’re a musician, music producer, or serious music lover, you may have already heard about the dreaded “LOUDNESS WARS” (start with Wikipedia, if nowhere else). In fact, maybe you’re even sick laready of reading about the subject. If not, then prepare for incoming rant.

Gather around the campfire, my children…

In the Old Days of music recording and audio engineering, engineers, having to deal with analog processing methods, were continuously looking for ways to minimize noise and distortion. This is because radio, vinyl and tape, being analog and subject to the laws of physics and chemistry, are inherently imperfect ways to transmit signals (i.e., music) and therefore inevitably a certain amount of noise had to be considered when working with these systems.

So over time, engineers and scientists developed strategies to minimize these noise and distortion components. This meant development of FM radio which was superior to AM for noise rejection, development of Dolby and other noise reduction techniques for magnetic tape, and implementation of a pre-emphasis curve for vinyl records. These were all clever verging on brilliant audio engineering techniques and resulted in a very high level of fidelity and noise reduction even before the advent of digital audio recording.

The early days of digital required some experimentation and scientific development having to do with word lengths, analog to digital converters, etc. But it very quickly became possible to create a recording in which the inherent noise was, for all intents and purposes, inaudible to the human ear and only measurable with laboratory equipment. The noise in recordings then became due to things like microphones, musicians touching their instruments or breathing into them, scraping their feet across the floor, etc.

So everyone should have been deliriously happy at that point, shouldn’t they? Well, by now, we know that’s not the way humans think…

Perfection, great! But not good enough…

Once people got over the shock/novelty of music recordings that had no clicks, pops, scratches, background noise, etc., CDs pretty quickly started to replace records. (The prices didn’t go down, in spite of the incredible drop in cost of manufacturing CDs, but that’s a rant for another time.) And people who listened to music at home, particularly classical, jazz, folk and quieter music were truly in bliss. Even rock music had more punch and power (if properly recorded and mastered) due to the greater possibilities of the CD medium.

The loudness race begins

But then…some bright spark decided that he wanted his band’s song to sound louder than all the other band’s songs that were playing on the radio at the time. That’s pretty easy to do, actually, using a technique called compression.

Basically, what you do is run your music through a device that essentially makes the quiet parts louder while leaving the loud parts untouched. It does this instantaneously and gradually. You can set it so its effect is almost imperceptible or extremely annoying (you can guess where this is going to end up.)

Compression has been used in recording since the beginning, both intentionally and unintentionally. In fact, most people can find completely uncompressed recordings more or less weak and distant to listen to. But like anything, this technique can be abused.

Producers quickly found that adding compression indeed made their songs pop out a bit more on the radio, and therefore get more attention and supposedly less station changing while they were playing. So other producers told their engineers that THEIR songs had to be EVEN louder. And the race was on.

By the 1980s and 1990s, the combination of extreme compression, combined with additional processing of the signal for radio broadcast and DJs whose normal tone of voice was a shout, resulted in FM radio stations sounding like basically they were at full volume at all times.

Unfortunately, this formula appeared to work. Stations were then all gradually sold off to one or two huge corporations, local DJs were fired and programs were created in large centralized studios, and playlists were pared to a handful of the most popular rap, country, rock, or oldies tunes depending on the stations format. And thus as far as I am concerned, radio died a long, long time ago.

OK, back to the subject. Apparently, it’s still important to some bands/producers/record companies to have the LOUDEST SONG. Remember too, that people are playing songs on all kinds of terrible playback devices: iPods, computer speakers,  handheld TV sets, car radios, club/disco systems, etc. This has always been the case, but there used to be at least a nod to the idea that many people had halfway decent “stereo systems” and would actually sit down in a room, put on a record, and listen to it. I don’t know if that sort of thing happens anywhere anymore.

Anyway, as producers got younger, and music styles changed (into rap, hip-hop, dance, etc.) the idea was that music should always sound LOUD. LOUD was GOOD. And LOUD mean compression and limiting.

For some styles of music, this actually does make sense. But for most music, dynamics are important. As an analogy, think of someone who only SHOUTS ALL THE TIME (think of the title of this page). Or a painter who only uses NEON COLORS [in fact, at this web page, a music producer showed what it would look like if master paintings were subjected to the equivalent of hypercompression]. We generally get tired of that person or that artist’s work pretty quickly. The human experience is supposed to have many dimensions of soft/loud, bright/dark, happy/sad, the spectrum of colors, sounds, etc. to interest us. Any time something just stays the same (like this page without subheaders to break it up) it becomes uninteresting and people move on.

Just two high profile offenders in the last several years are Californication by the Red Hot Chili Peppers and Death Magnetic by Metallica. However, there are countless others which you can find with a little searching.

The Death Magnetic case

Death Magnetic is a very interesting case since there has been a lot written about it in forums and magazines. Opinions range from “it’s mixed so brutally that it’s unlistenable” to “hey, it’s what heavy metal sounds like, either deal with it or don’t listen to it.” I bought some copies of it and tried to do some measurements on it to attempt to actually get some facts to argue from.

In a nutshell, this is what I found out. The commercial CD release (and the MP3 or AAC encoded version you can buy online) are, for some reason, compressed and brickwall limited to a  point of ridiculousness. The dynamic range averages about 7 dB, whereas some of the heaviest and most powerful CDs of the last 5 years range from 9 to 15 dB (the way I measure it). Looking at the waveforms close up in an audio editor shows that most of the songs consist of clipped square waves. There are probably other CDs as badly mastered as this, but very few major releases. The vinyl release doesn’t sound much better, and the dynamic range is only slightly larger (about 11 dB).

I then read some rumors online that there was a version of the album in the video game “Guitar Hero” that had all the songs, but mixed completely differently with much more dynamics than the commercial release, and I was interested in seeing whether that was true.

I located a copy of the Guitar Hero version (GH) and submitted it to the same measurements. Astonishingly, the same tracks came in at about 15 dB dynamic range. On top of that, my subjective opinion is that the GH version sounds outstanding, far, far superior to the commercial release…

I personally never listen to my regular released copy of the album but only the “GH” version.

I even bought the two-disk vinyl record release. It was apparently mastered from the same version that went on the CD; and is nearly as unlistenable. (Anybody want to buy my copy for a low price?)

There’s an engineer named Ian Shepherd who has really analyzed this issue in great depth. His blog is a great resource to read all about the loudness wars, their current state and whether they can be “won.” Here are a couple of links that will get you to the discussion specifically about Death Magnetic:



Digression: how I estimate dynamic range

I do realize that the way I do this is not completely scientifically accurate, but it does give me a measurement that allows me to compare tracks and passages.

Basically, what I do is first normalize the audio I want to analyze to 0 dBFS. Then I compute the RMS level of the section (I use the Statistics tool in Sony Media Works Sound Forge to do this.). That gives a measure of the average energy in the section. The higher it is, the louder the section is. Looking at this number for many, many tracks, the inflation of average value for pop and rock music has clearly gone up from roughly -20 dB to -12 dB over the last 30 years or so.

A much better technical discussion, and a fascinating essay about how radio broadcast compressors work, can be found in Mastering Audio by Bob Katz. (Incidentally this is an essential book for anyone interested in audio recording and processing.)

Nothing could be louder than Slayer! Right?

I did an interesting experiment. I took my Reign in Blood CD and ripped it to audio files. Amazingly, the peak levels of the files were actually 2 dB below full scale! This means that the CD could have been made 2 dB LOUDER without requiring ANY processing at all! Apparently, whoever did the transfer from the original analog stereo tapes to the digital CD master either was too conservative or just made a mistake. But adjusting for the 2 dB (normalizing), the CD STILL only averages about -15 dB, even though it sounds intensely powerful and loud.

Recent (Mar 2009) links, discussion and blog entries

I’m sure you can use google as well or better than I, but I did want to put some links from recent readings on this subject here. I can’t say I have much in the way of analysis or discussion of them, but I want to at least get them listed for reference…

from Ian Shepherd’s Mastering Media blog

John Scrip on Death Magnetic from his Massive Mastering blog

Music Producer’s Guild (UK) press release about loudness war roundtable discussion (Mar 2009)

(This is really a good place to go for great discussion about Loudness Wars in general, and the Death Magnetic case specifically.)


A stab at fixing the dynamic range problem? “Pleasurize Music

(Oct 7 2009) This outfit has apparently been around for some months, but I just learned about it. They’re trying to create a critical mass of producers, engineers, artists, music fans, etc. and create some standards and labels to indicate the dynamic range of records.
(Aug 26 2011) Hello? Hello? Is this thing on? Haven’t heard anything from this outfit. Hope people haven’t been sending money.

Page#73/last edited 20110925