Audio Mastering Made Simple (not)

(This is a rough draft. The post will evolve. I'll add some illustrations and maybe even a video, but for right now it's just the basics.)

JARGON ALERT: If you don't understand the jargon used in this blog entry, refer to this glossary of sound engineering terms.
---------------------------------
So I've got a few decent mixes that I'm making radio-ready. That means mastering, the part of audio engineering that separates the adults from the children, the pros from the amateurs. It's the part of the process where your mix gets big and sweet.

It's also the hardest part. It takes hours and hours to do one song properly. Worth it, though.

In the age of digital music and digital mastering, the true art of it seems to be getting lost. If you don't believe me, spend an hour really listening to anything recorded and mastered before 1983. Ideally, dig up an LP on wax if you can. If not, get a CD that doesn't say "Digitally Remastered!" on it, as if that were some sort of good thing.

If you don't have a turntable with a real needle for finding analog vinyl grooves, you are forgiven. Not many people do. I don't at the moment, actually.

Just listen to the old stuff for an hour or so. Then listen to the new stuff. If you have a keen ear, you will notice a lot of things.

In the beginning, there was the input...

First you will notice that the old stuff is "groovier" and more human. This is because actual human beings played the songs on actual instruments in an actual studio. Generally, they played the songs all the way through. They might have done it track-by-track, and there was certainly some punching, but every track was played all the way through. Punching in the pre-MIDI age was a major PITA, so both musicians and engineers tended to avoid it.

Nowadays, by contrast, the ease of punching in and out has made the whole process into one big punch-fest. Instead of playing the whole song, they generally play the parts of the song, at least in the commercial world. They play them over and over. When you hear a superstar recording artist singing a song on the radio, they didn't actually sing it like that. They sang the chorus about 500 times. They sang the verses about 500 times. They sang individual lines and even words over and over. The engineer took the best 20 takes or so, layered them, cross-faded them, compressed them, slathered on a little reverb or some other effect, and -- voila! -- perfection. The same goes for every other track and every other instrument.

And if you're wondering why a live concert so often sounds just like the album, it's because the superstar recording artist and their band are basically lip-synching (it's not actually that simple, but that's the long and short of it). You didn't really think that chick (Jessica Simpson's sister, was it?) who got caught with her pants down on SNL a couple years ago was the only one, did you?

Anyway, the input was different back then. The input is important to the final finished master, obviously, which is why I mention it.

Then there's the mixing...

In my opinion, there isn't a whole lot of difference at this stage -- it's just different tools being used. In the old days, they lined up all the tracks and played the tapes simultaneously. A mixing board with faders is basically the same thing as a digital audio workstation like ProTools. Of course, nowadays you don't need $50,000 worth of effects processors to tweak the sounds since all of that is built right in to the software so that's kind of nice.

Here's what a million wannabe rock stars ask themselves a thousand times every day...

Why is my mix so flat and why isn't it as loud and full as the stuff on the radio?

Yeah, you used ProTools or ACID or whatever, and it sounds great. But... it's just not there, ya' know?

This is where mastering comes in. To master, you have to understand sound at a fundamental level, at the level of waves and frequencies. You have to understand that the human ear hears a nearly infinite range of frequencies and that some frequencies are outside the human range. Gaining fine-grained control over these frequencies as they exist in your mix is what mastering is all about.

To do this, there are three tools: EQ, compression, and limiting. Before I talk about these in particular, I need to tell you about something very important: headroom.

It all starts with the headroom

"Headroom" means giving your mix some space on top so that you can boost the frequencies that you want to boost. If your mix is too loud, you won't be able to make the bass bassier or the screaming guitar screamier without distorting everything and making it sound like crap. Turning down your input from the mixer to the mastering software is a totally amateurish way to gain headroom. Your master will not enhance your mix much if you do it this way. Just a caveat. You have to do it the hard way: Spectral analysis.

Spectral analysis...

... is a fancy way of saying you need to get a visual of the frequencies that you're about to EQ before you EQ them. Note the ones that can be nudged down. 1.5 kHz poking up like a dandelion on the lawn of the foreclosed-on house across the street? Tamp it down. That's the first function of EQ -- getting maximum headroom without sacrificing volume.

What the hell is EQ?

EQ is equalization. It's the means by which you shape the frequencies of sound mentioned above to maximize the desirable ones and tone down the undesirable ones. It's super duper important.

Every decent stereo system has a rudimentary EQ system, usually 5-10 bands. To master audio professionally, you need a minimum of 30 bands. The software I use has an HDEQ functionality with 1023 bands. (I don't use it because 30 bands are enough if you know what you're doing.)

Some audio engineers are coldly scientific about EQ. A particular frequency should be boosted for richer bass, another should be faded down to reduce hiss (and the "S" and "F" sounds -- it's called "de-essing," and it's why you'll hear roadies say "sibilants" into the microphone during sound checks sometimes). No engineer will tell you that one EQ scheme fits all, though. A great deal of proper EQ has to do with the input, and even more of it has to do with the effect you're going for. The same recording can produce either a thick, muddy sound or a bright sound depending on the EQ.

When mastering, you break your 30 or more bands into three basic ranges: low, medium, and high. If you're using professional software, you will be able to define the limits that separate these. You should be able to mute or solo each range to gain the super-granular access to the sound that you'll need. (More on these three basic ranges when I talk about compression.)

OK, let's talk about compression...

I'll be honest with you: For the most part, I hate compression. Well, that's harsh. I don't hate compression but I hate the fact that it's over-used and that in the past 15 years or so its overuse has largely slaughtered one of the most sublime sources of nuance in musical performance and composition: dynamic range.

When human beings play music, the volume of each sound varies. This is called dynamic range. When I play my guitar, I play certain notes loud, mute others, tap my fingers on the body of the guitar, etc. When I sing, sometimes I want my voice to trail off or jump out -- even within the same song. Thanks to the magic of compression, dynamic range is lost in commercial music.

Listen to the radio if you don't believe me -- every sound in the mix is essentially the same volume, especially on the radio because the radio station compresses the already over-compressed recording further. They do this because compression is the easiest way to make it loud overall without distorting the sound.

If you listen to Led Zep on wax, you will have a different experience than the one you get on classic rock radio. Trust me on this.

Anyway, compression is an important part of mastering. The traditional way of doing compression was to have three compressors, one for the low range, one of the medium range, and one of the high range. Most pro-grade mastering software is set up this way as well. If you want a bright sound, you might compress the hell out of the high range and leave the low range more or less alone. If you want a muddy sound, you might do the opposite. This is just one example of how compression is used. You really have to play with it -- but please don't overdo it, at least if you want old farts like me who enjoy dynamic range in music to listen to your stuff, which you probably don't... but anyway...

Limiting -- turning it up

So now you've got your mix sounding bad ass. The last step to making it fully professional is to define the limiter input. This is easy -- just turn it up until the output channels are bouncing around near 0.0 and the sound isn't distorted. I like to push it up until it is distorted, then notch it down until it's not -- and not a 100th of a decibel less. If you got the sound you want, loud is good, right?

In conclusion...

Mastering is really important. :)

PS -- I'll add more to this post soon -- including graphics and some audio samples of mastered vs. unmastered recordings. So check back. Just had to lay it down for now.