How Compression and Normalization Affect the Listening Experience

One of my biggest complaints with CDs isn’t a fault in the media; it is the fault of the idiots who master them. It seems there is but one goal for them: to ruin my hearing. Great sound and great music require there to be several elements: punch, definition, and clarity.

Think of some of the most identifiable natural sounds. The crack of lightning followed by the roll of thunder, the spring peepers croaking in the woods, and the roar of the ocean; these sounds have a melodic quality, a rhythm and definition, and a definite change in decibels as the sounds are heard. Sounds that are considered pleasant most often mimic the qualities of these natural sounds.

There is a class of guys who lay down soundtracks for movies who understand this. Music becomes part of the experience: booming and fast paced in action scenes, soft and subtle during romantic scenes, and so on. This is demonstrated in movies like Lord Of The Rings and Legend where the music is blended in so well you almost don’t know it’s there. There are, of course, exceptions to this — movies like Mystic River where the music track completely drowns out the voice track at times.

Most music CDs don’t seem to get the kind of attention they should. Music CDs are compressed — a term that has nothing to do with the amount of size a file takes up, but rather the amount of range in the volume of a piece of content. By compressing audio, you never end up with sounds that are too quiet to be inaudible or so loud as to be deafening. Instead, you end up with audio that is very monotone.

This image shows four audio tracks. The first is an example of how audio should look, with a great deal of change in volume within the file: occasionally being truly silent, and occasionally being at its peak.

Next is an acceptable example. While peak isn’t reached at any point, there is a good amount of variation in the volume.

Sample three suffers from severe compression. This is a piece that no longer has any diversity — it is all monotone. This piece also never reaches a peak level, so to make the song soft, it was compressed but not normalized.

Sample four is very close to perfect. There’s good variation in volume, though it is normalized to a bit beyond peak (which can cause distortion).

In CDs you buy today, sample three is the most typical. The audio is compressed to the point where there is added distortion and all the “life” is gone. The music becomes severely ugly. This, of course, helps sell concert tickets because these days most everyone does sound better on stage. It also helps MTV as the unplugged music sounds much more lifelike than the music that gets passed off on CD.

Going back to the thunder example, below is a WAV of a thunder clap. The green is the WAV before audio compression, and the red is after. (I used Adobe Audition’s Radio Compression preset to ruin this beautiful sound.)


Compare the two mp3s. This demonstration is quite dramatic because the roll of the thunder (in the compressed example) doesn’t get softer the way it ought to. This isn’t to say that compression has no purpose. It is great for interviews and spoken word content where you want to keep mumbles audible and shouts to a small roar. But compression takes a lot of the emotion out of the sound. It’s like the difference between listening to NPR or listening to Martin Luther King, Jr.: an audio compressor makes everyone sound like Lynn Neary and Robert Siegel.

Because online music such as iTunes and Napster 2.0 use CDs as their source, the music I buy for my portable device sucks, too. Only it sucks more because instead of getting CD quality suck, I get near-CD quality suck. The distortion and loss of frequency response that results from a file being data compressed / encoded to 128 or even 192k stacks on top of the distortion added via audio compression.

Because each format encodes slightly differently, some content will sound better from one online provider than from another. AAC tends to handle content that is audio compressed better than WMA does, but WMA tends to handle the softer and more subtle sounds found in uncompressed audio. Because so much of what is sold online is pop music, this puts AAC in a better light than WMA in the format wars. If you are, however, interested in preserving your RCA Red Label 1812 Overture or you’re just transferring it to your portable media player, might I recommend WMA? You will find the cannons’ rumbles to be crisper and the subtle sounds of the children’s choir fading in and out to be much cleaner in WMA. Mp3 will lose some of the stereo separation that is much more important in orchestral music than it is in pop. If you look at the examples above you will notice that most of the music is practically monophonic; the audio in the left and right is almost identical, whereas the thunder is very different between channels. AAC has stereo separation equal to that of WMA but it tends to drop sounds below a certain db. Whether this was done as a noise reduction feature or if it is just a way to drop some of the bits that would never be heard by someone who has been to too many concerts, I don’t know.

I’m 32, so I don’t remember tube amps or 8 tracks, but I often think maybe it would be better to roll back to those days. When I bought my home theater, the salesman was trying to push me into a model that included 20+ surround modes from living room to concert hall to cave. Sure, that is cool for my Audigy card when I’m playing games and the programmers don’t have to presample all of those environments for each sound in the game, but like I would want to watch a movie in a cave, or hear Britney as though she were performing at the bottom of the Grand Canyon. (Okay, Britney and Paris Hilton in the bathroom surround mode crossed my mind briefly as I figured I could pretend they were naked in my shower… but that passed pretty quickly.) Perhaps I’m too much of a purist, or perhaps labels think their target audience is the guy with a 4000-watt, 42-inch subwoofer in his car. But for me, I’d like to hear the music the way you would in the studio — with all the subtlety and life still in it.