Audio Class With Professor Hoffman:

Lesson 3: Compression

Can you help us (non-studio drones) understand compression better? As I understand it, you *need* some compression, because you don't want a trumpet solo (for example) way louder than everything else. So it's necessary, correct? 

Well, compression is the most misunderstood technique in the recording process. I could write a book on its uses and misuses. 

Let's see. 

Yes, we need some compression in our music. The sounds of real life won't do for most recordings. There has to be a way to make everything simmer together, like a good Italian sauce. But if you overcook, or overflavor: disaster! 

Easy instruments to record? Well, a guitar amp works as a compressor too. The harder it is hit, the more compressed the sound; nice tube overload. The volume remains constant, no matter how loud it gets. That makes the electric guitar one of the easiest instruments to record. The human voice is one of the hardest. I'm sure you have tried recording your voice into a tape recorder, watching the VU meters go as you do it. If you sing too loud, the tape goes into total distortion. So, you have to turn down your microphone. If you sing too quiet, no one can hear you at all. Our voice has about a 90db or more dynamic range. Your VU meters only have 20 or so db to measure, and most pop records of the 1970's have about 10db dynamic range. So, the voice has to be "limited" when recording, or else the quiet notes will vanish in the music, and the loud notes will overpower everything else. It's tricky to get it just right. 

You wanted a few examples of two different versions of the same song with different compression ratios? OK. Take the case of Pet Sounds. Listen to the mono mix of (let's say) "Wouldn't It Be Nice". A lot of warm tube compression keeps the voices "in the mix", riding along with the music track. Not the best mix on earth, but totally of it's time. Now, listen to the stereo remix of the same song, ignoring the fact that it seems more open because of the stereo effect. Just listen for tonality and mixing style. The voices have a squashed, edgy quality that make them stand out too much. They are overcompressed, and then EQ'd to make them more "transparent". Problem is, they don't flow with the music anymore. They sound like they were recorded in a different time zone then the music.... 

Another example. The mono version of "Beatles For Sale". Listen to "Eight Days A Week" on the CD. Now that is overcompressed! The poor limiter has hit the wall and can't compress anymore, so it just distorts. Ouch...Overload! The stereo version of the song was mixed with much less compression, using the famous Fairchild stereo tube limiter (used on all stereo Beatles mixes). This version sounds much better, and it's not just because it's in stereo. Even played back in mono, the stereo version at least breathes a little, and since the limiter is not bottoming out, the song has a nice 10db of dynamic range. Pretty nice for Abbey Road in those days. On the other hand, the mono mix of "Eight Days A Week" has about 2db of dynamic range. Youch! Even the voice on your telephone has about 25db of dynamic range! More than any Beatles mix... 

Now the Byrds recordings sound like that because that is how the engineer and producer wanted them to sound. Mainly EQ choices, with a lot of limiting, especially on Jim's (Roger's) 12 string. That sound can give one a headache after a time, eh? 

Compression is what makes something sound louder, but really, it just removes anything softer. I can whack a snare drum, and the microphone will overload, and the vu meter will pin. Just one snare whack, and the whole recording is ruined. Why? It's really loud and dynamic. But, with compression, it can SEEM loud, but it really won't be. Get it? It's like when you watch TV. The shows seem normal in volume, so you turn up the sound to hear everything clearly, but when the compressed commercial hits, you cover your ears. But, there is a point in the show where the volume actually hits THE SAME dynamic peak as the commercial. It's just that the commercial STAYS AT THAT VOLUME ALL OF THE TIME, whereas the show only hits that peak when a person is yelling or something... See? 

Finally, think of compression visually like this. You are standing on one side of a sliding glass door. Someone is on the other side, and as you watch, starts pushing their face against the glass. The face doesn't get any closer to you, it just starts to look squashed, like a good 90mm camera lens will do. You don't want the person's nose to look really long and unnatural, see? You want the perspective to be "flattened" so it flatters the person's face. Well, same with music.