FIX YOUR MIX .com » BLOG

I got an email from my father today directing me to an internet forum debate regarding how over-compression in mastering is ruining music today and I responded with what you’re about to read–the simplest, most easy-to-read breakdown on this topic I could come up with–an explanation that anyone will be able to understand. No flowery or abstract adjectives, just the meat and potatoes. If something still isn’t clear, leave a comment. I’m mostly posting this article to save myself some time (to direct our clients to) but this post may be helpful for other mastering engineers who’d like to do the same.

Note: this post refers to sound compression, not MP3 or zip compression. Learn more about the differences between sound compression and data compression here.

What Is Mastering Compression?

Mastering compression* (”limiting”) reduces the dynamic range of a recording. Dynamic range is the range between the loudest and quietest point in a recording or section of a recording. Dynamics and loudness are inversely proportional: the less dynamic a recording is, the higher the average loudness of that recording can be, and vice versa. Today, there is said to be a ‘Loudness War’ between labels trying to release the loudest records possible.

The Pros and Cons of Loudness

When we’re talking about the consequences of loudness, we’re really talking about two different things:

• From a macro perspective: A louder/less dynamic recording means all of the sections of the song will be about the same volume. For ‘Verse Chorus Verse’ style songs, the benefit is that the song comes in loud right off the bat and stays loud from section to section. The downside is it means the chorus doesn’t “hit you” or sound as big as it otherwise might because there is little or no change from the verse to the chorus. In fact, if there’s more stuff going on during the chorus, individual elements may actually get smaller. For example, heavily compressed rock mixes tend to have a bigger snare sound during the verses than the choruses.
• From a micro perspective: Compression from one beat to another is hard for the untrained ear to hear, unless it’s very extreme. And even then it is hard to explain what it is you’re hearing–you just know it sounds bad. At this scale, compression makes the mix sound more “exciting” right away, but if overdone can be fatiguing on the ear to listen to. Perhaps more importantly, the drums will often be less punchy if a mix is more heavily compressed.** A former coworker and assistant to one of my all-time favorite mixers once said to me, “mastering ruins everything.”

Is Louder Better?

There was a psychological study which showed that people consistently prefer recordings that are louder, even by an increment as small as 1 dB, even when they’re not told what the change is. (Anyone want to find the link for me?) If you want an extreme example of over-compression, just listen to the radio. They use a more complex system of compression to get recordings even louder than CDs. And yet for every person who puts up a fuss in the blogosphere/messageboardiverse about mastering ruining music today, I have a memory of someone in my childhood telling me how they like the sound of radio. It just has that magic je ne sais quoi. There have been studies which indicate that loudness has a strong effect on which radio station a person will stop on when channel surfing.

While I consider myself to be more of a ‘new school’ engineer rather than one pining for the days of yore, I sometimes wish more records today had bigger dynamic changes. My favorite example is “Quiet” by the Smashing Pumpkins, which came out in 1993 before the Loudness Wars really began. It will probably be hard to tell on YouTube, but when the guitar solo comes in on the record, the song just gets so much louder. If you already have the song cranked, the solo will hurt your ears a little bit. Now that, ladies and gentlemen, is rock & roll.

For a case study in consumer backlash against loudness, check out my post about the Metallica album Death Magnetic.

*The type of compression I’m talking about here is “limiting”, a specific type of compression that comes at the end of every mastering engineer’s signal chain. I’m using the term compression throughout the post because it will be more familiar to musicians and readers.

**If I know that a project I’m mixing will be mastered by someone else, I usually try to make my drums a little punchier than I want them in order to compensate for the effect that mastering will have, unless I know the mastering engineer tends to go easy on the compression. This is also why it is usually best to select a mastering engineer that your mix engineer has worked with in the past, so that the mix engineer can anticipate what will happen to the mix in the mastering stage and mix accordingly.

I knew it was going to be loud.  I think I knew it intuitively just from the clues hidden in the mix of their timeless album Loveless.  I definitely knew it before I heard people in the parking lot spreading apocryphal stories about audience lawsuits over hearing loss.

So before we arrived at the Palladium Ballroom to see My Bloody Valentine on their 2009 reunion tour, I decided to be a good friend and buy a big jug of earplugs for my group of audiophiles in case they forgot to bring any.  Sure enough they did.

It turns out that it didn’t matter because as we walked through the gate, the ushers handed each and every person their own individually packaged set of earplugs.  That’s how you know a band is going to be loud—when the venue makes a special effort to put a set of plugs in every concertgoer’s hand before the show.  I’m used to seeing a jar of cheap plugs at the bar, but this was a clearly a very different animal.

It was ungodly loud.  Perhaps the loudest thing I had ever experienced.  My dad was an airline mechanic and on Bring Your Son to Work Day I got to experience just how loud a jet engine is.  My Bloody Valentine was louder.

During their closer, an extended rendition of “You Made Me Realize”, the band chose a single chord and extended it for 17 minutes (I know because the person in front of me was recording it and had a timer on the LCD screen).  There were people wearing earplugs still trying with all their might to cover their ears with their hands.  There were people doubled over in pain.  Frankly, I couldn’t believe the sound system at the venue didn’t explode.

The experience alone made the show worthwhile, not to mention getting the chance to see one of my all-time favorite bands.  But there were a few things about the show that made me feel cheated.  If you see MBV play live, you’ll be sad to see that there is no synthesizer player.  All those great synth hooks are pre-recorded to a tape.  Without those melodies, you are pretty much left with two people playing rhythm guitar, a bass player, and a drummer.

Also, the vocals were buried to the point that sometimes I questioned whether or not anyone was actually singing (You might also wonder that from the decidedly un-engaging stage presence of the seminal shoe-gazers who seemed to exchange staring at the floor for hiding behind the microphone). In the world of live sound it’s almost cheating to hide the vocals that way—it is very easy to make a band loud if you don’t have to worry about feedback (or even hearing the vocals for that matter).

But to a certain extent that’s what Loveless sounds like.  It is a lush and thick wash of guitars with soaring synth lines against buried vocals and drums.  All this contributes to a very unique sonic texture that has extremely high average loudness, but subconsciously forces the listener to turn up the volume.

Prince famously said that the most important part of a mix is the “boom and the slap” (meaning the kick and snare).  Well, in Loveless, the drums are mixed so far back into the track that at normal listening levels they are barely noticeable.  In order to get the drums to a listenable level, the listener is forced to crank up the whole track.  By the time you get the drums to where you want to hear them, the unwavering assault of distorted electric guitars is screamingly loud.

Since I cut my teeth in the music world as a drummer, the drums are what I notice.  But many people site the same experience by trying to bring the vocals to a listenable level as well.  They pose the same problem, although to me the problem of audibility in the vocals is not remedied by additional volume since they are so legato, washed in reverb, and poorly enunciated.  But then again, lyrical content and vocal execution weren’t the primary objectives.  The point was to be loud.

For the most part, the synth hooks and even the snare drum cut by frequency content rather than relative volume in the mix.  They occupy higher registers sailing over the bed of guitars.

From a mastering standpoint, the drums and vocals have to be mixed this way so that they become part of the texture. If the transient content of the drums were any louder, the track would pump wildly as the compressor keyed on each drum hit.  Similarly, the vocals couldn’t stand out like most lead vocals do or else the mastering compressor would key on it instead as well.  The overall sound is accomplished by creating a bed that has very little fluctuation in average loudness despite any changes that might occur.

Of course I’m sure that none of this was explicitly sought after.  Like most indie bands, MBV probably tucked the vocals in because they didn’t like the sound of their own voice.  But nevertheless, the end is achieved masterfully, regardless of the means or the motivation.

I love the way Loveless sounds, but I will be the first to say that that record sounds terrible if it isn’t cranked way up.  The mix really only sounds reasonable if the volume is up, but that’s all right for pretty much anybody who wants to listen to that record anyway.

My girlfriend and I met at a Swell Season concert.  Many of you might know them as “that band in the movie Once” that won an Academy Award for Best Original Song from a Motion Picture, but they are in fact a real couple with a real band making real music.

Our anniversary came up on Tuesday (she’s put up with me for a whole year now) and we were taking a trip down memory lane.  Being sentimental types, we kept trying to find ways to incorporate the Swell Season into the evening but couldn’t quite come up with anything exciting beyond listening to the record (which we didn’t end up doing anyway).

The band is great live.  Glen Hansard is a bit chatty on stage, but it is an engrossing couple hours of storytelling (and they’ll start touring in November so get tickets if you can).  But the film is fantastic if you’ve never seen it.  It probably has the best making-a-record sequence committed to film.

The only thing that bugged me about that act in the movie was after they finished recording, the engineer turns to the group and says “Well, let’s give it the car listen” and they proceed into a very scenic montage of the group piling into a car and driving around Dublin listening to the record.

The idea is that no matter what, a professional CD should sound good in the car.  It doesn’t matter who mixed it or produced it or what kind of car you drive, you should NEVER put a commercial album into your car stereo and say this doesn’t sound right.

In practice this is totally false.  In fact there are a number of CDs that sound absolutely horrendous in the car—The Who’s 2006 release Endless Wire comes to mind (which is a shame because Pete Townsend takes the time to outline every piece of vintage gear that they recorded and mixed with in the liner notes; proving once again it’s not what you have, it’s how you use it).  But still, the car listen is a pretty decent barometer in most cases.

Personally, I hate the car listen.  In the eyes of many pros, the car listen demonstrates that you aren’t familiar enough with your room and your gear to know when you are making the right decisions.  For most situations, you should be terrified if you are in the studio with a house engineer who says “Well, let’s give it a listen out in the car.”  That’s a bit like an architectural engineer saying “Well, I’m pretty sure the building is level, but we may want to give it another look around noontime when the light is better.”

Admittedly there are times when it might be necessary to use the car listen.  When you first come to a new studio, you might take a book of CDs into the new listening environment and play them in the control room to get used to the character of the space and the gear.  Then you might take your first mix out from the control room to the car, which is a portable, familiar listening space that has reasonable frequency range (as opposed to headphones), and test the mix out in order to make sure you properly assessed the room and made the right adjustments.

So what makes the car listen so appealing?  Well there are a number of factors that are present in automobile listening situations that you can’t test in the somewhat academic environment of a professional (or semi-professional) recording studio.

For one thing, everyone is familiar with what their car sounds like, so clients like to hear things in the car because they don’t have to actually go home to hear their music on a familiar loudspeaker system.  They can just walk out to the parking lot.  Of course if you go into the engineer’s car (like they did in Once) that is only for the benefit of the engineer.

Another benefit is that you get to hear what the record sounds like in noisy environments.  Studios are meant to be clinical, noiseless environments where you can hear minute details and imperfections.  They are ideal listening situations, but the real world is rarely ideal.  Most of the time consumers are going to listen over the hum of an engine going 60 miles per hour with the AC blaring and wind gusting against the windshield.

In those instances, you will lose most of your top end since environmental noise buries those frequencies very easily.  Also, bass reproduction suffers in the car because much of the energy goes out away from the car as well as inside it.  The thin material of the car doesn’t retain and resonate low frequencies the way it does mids and highs.

You can listen to the record in a real world situation and say “Well, Kanye usually gives me a bit more low end than this on the highway” and adjust accordingly.

Bear in mind that I’m saying a professional engineer who works at a particular studio consistently should never suggest a car listen, if you are patronizing a studio though, by all means take the record out to the car and give it a listen.  The engineer might even be grateful that you broached the subject so he doesn’t appear unprofessional. 

At the studio, you have to trust that the engineer is making the right decisions with your art, but it is never a mistake to check his work.