Peak Limiting, Loudness Wars and Remote Mixing

Using the fine adjustment tool to crush the mix.

You probably heard of loudness wars by now. Back in the ’90s, somebody thought it would be a good idea to make a mastered mix as loud as possible so that it would sound explosive on the radio. The irony is that, on radio, quieter classic records from the 70s often sound bigger than today’s aggressively limited joints.

Wait a minute—that makes no sense! But it’s true. If during your next radio listen you happen to hear a finger-picked James Taylor tune like “Mexico” bookended by virtually any two heavy modern rock songs, you’ll be amazed at how rich, punchy and detailed it sounds relative to the others. Same thing goes for Joe Walsh’s “Life’s Been Good” or later Led Zeppelin records. If you really want to be blown away, hope that you get to hear Bob Clearmountain’s mixes of “Weather With You” or “It’s Only Natural” from the 1991 Crowded House album, Woodface, or his Beatle-esque tour de force, “Sowing The Seeds Of Love” by Tears for Fears. Those FM radio staples virtually jumped right out of the speakers when they were freshly minted, and they still shine today.

So, why would a quieter record sound better than a loud one, on the radio? Answer: dynamic range.

Radio compressors and limiters are designed to be a one size fits all answer to the problem of over modulation, or “overs” that you’ve seen when digital peak meters go into the red, indicating clipping. If you were a radio station, you would not want overs because they would cause the FCC to fine you. If you were a record company or a vinyl pressing plant, you would not want them because they might literally cause the needle to jump right out of the groove, especially on a bass-heavy song!

Anyway, radio compressors and limiters sound more musical when they can react to a decent amount of dynamic range, of which modern rock records tend to have very little. Super loud, overly crushed records drive the compressor into gain reduction, where it frequently remains until the quiet breakdown or the end of the song. One of the symptoms this is a strident, distorted sound that fatigues the listener.

Okay, but what does this have to do with remote mixing?

The exact same mix, played twice with a 1 dB (or less) volume difference, will always cause the listener to believe that the louder one is better. What this implies is that my colleagues and I need to make our final mixes at least as loud as the reference mixes we receive from the artist. If we don’t do this, we lose. Nobody wants to pay thousands of dollars for a professional mix that does not sound as “good” as the band’s rough mix.

Of course, the pro mix will always sound better if the listening environment is accurate and the two mixes are level matched. These factors are easily controlled when the artist attends the mix session. Plus we discuss artistic goals, decisions are made together, and the process is transparent. The artist becomes an invested partner in the mix. When we compare the rough or reference mix to the new one, we are always doing so with matched volume levels. It’s super easy to demonstrate the difference in audible distortion, so the artist buys into the idea of clarity and punchiness instead of sheer loudness.

When mixing remotely, however, it’s difficult to guide the artist towards an optimal listening environment. The first thing the artist tends to do is to load the new mix into the original Pro Tools session to toggle an A/B comparison of the new versus the old. This can lead to two potentially huge problems. First, artists and bands typically do not know how to tastefully peak limit a mix, so, on a quest for maximum loudness, they squash the life out of their rough mixes to the point that static instrument levels begin to change over the course of the song, depending on how much gain reduction is being applied. If I don’t want to lose the battle of first impressions, I must compete with this loudness. Second, the artist may in fact be listening to my mix through their overcooked compression and limiting chain because they may forget to deactivate it from the mix buss.

Fortunately, experienced artists and repeat customers have already been down this road, so they tend to leave the peak limiter off their rough mixes. They trust me, because we’ve been successful together in the past, or they want to attain the sound of my records that they already know and love. New artists, however, are looking for any excuse to go the DIY, do it yourself, route. They do not want to spend money on crafting their records because they do not believe that they will ever make any money from sales of those records. They expect to hear maximum loudness on the first listen so they can compete on loudness instead of artistry. For those guys, I add a peak limiter to the first MP3 that I email them. I make the file the same loudness as their rough mix to dazzle them and earn their trust, then I ask them if I can take it off and do my thing. After I assure them that the final master will be at least as loud as the reference MP3, they’re cool with turning up the volume knob to listen loud.

It’s funny to me that as a guy with audiophile tastes, I sometimes prefer the energy and sound of limited masters to the (relatively) pristine clarity of the mixes. In order to tell with certainty, however, the two audio files must be level matched. Creative use of peak limiting can add a sonic crunch to rock songs that creates a sense of urgency. Limiting is almost required if you want to ensure that your record will be audible over the engine and road noise on your car’s stereo system. The unmastered mix will sound terrific in a neutral acoustic space, but mastering will ensure transportable translatability and competitive loudness. Peak limiting is a necessary evil that isn’t necessarily evil if done with prudence and good taste.

With that said, I wish all my artists and clients could be present for their mix sessions so that they could hear the dynamic beauty of their song in its pristine unlimited glory at least once. The experience might give them the confidence and courage to avoid getting sucked into the loudness wars. It’s ironic that they go to such great lengths to ensure that every detail is audible in the mix, but then they allow and encourage everything to be steamrolled during the mastering process— especially ironic considering that a reasonably dynamic mix will sound bigger on the radio than a squashed one!

Kathleen Wirt (right) and I are happy because Marek Stycos fed us Thai food after evangelizing about audio quality. Kathleen and her crew at 4th Street Recording in Santa Monica do things right. Great vibe there!

Mix Tip: Lead Vocal Compression/Limiting  


Happy singers = happy mix engineers!
Do you ever wish you could make a lead vocal sound “urgent” and heavily compressed in the mix, without all the nasty pumping artifacts? We’ve all hyper-spanked a vocal to create a glorious larger-than-life character because it sounds uber cool to do so, only to find that the sparkle disappears and the breaths become almost louder than the words. 

Despair not, for I have a technique for you to try!

Instead of driving a single compressor/limiter into super heavy gain reduction, try using two separate compressors in series. Keep the gain reduction small, and the ratios low. 

A classic sound is Teletronix LA-2 or Urei LA-3A into a Urei 1176. Opto into FET. I guess the theory is that the opto compressor smoothes the dynamics and shapes the tone, then the FET further limits the dynamic range and adds groovy harmonic distortion. 

Generally (but not always) I prefer going the opposite direction. I use the 1176LN to precondition the signal before it gets to the the opto. This is a cool sound. Preconditioning the signal ensures that rogue peaks do not cause the opto to go so deeply into GR (gain reduction) that the recovery (release) becomes too slow for the rhythmic cadence of the melody. We don’t want the recovery so slow that it steps on the beginning of the next vocal phrase. 

By putting the fast FET limiter first, and setting it to get in and out quickly, we can keep the opto in its sweet spot. If you’re not sure where to begin with your 1176LN settings, start here: input 10 o’clock; output 2 o’clock; attack 10 o’clock; release 2 o’clock; ratio 4:1. These input and output settings, of course, are based on the assumption that the source signal was recorded at a proper level for a “+4” analog world, where pro audio equipment is designed to work in its sweet spot. 

If your taste is anything like mine, you will quickly find that the 10 o’clock attack setting is too fast and the 2 o’clock release is too slow for modern rock and pop vocals. More often than not, I crank the attack all the way counterclockwise (the slowest setting)  to allow sparkly transients to pass through, and I slam the release fully clockwise (fastest) before carefully twisting the knob back the other way to make it incrementally slower until the pumping disappears. 

Faster release times make the signal seem to be louder, more urgent, and more aggressive, but not as smooth and polished. Experiment with this by closing your eyes and listening to what happens as you turn the release knob. Superfast release times will sound almost brittle if they are too fast.  It’s an artistic choice, not a matter of right or wrong.

Circling back to the input and output settings, in this series configuration I like the sound of the 1176 when the VU meter is bouncing around from zero dB to minus 3-4 dB GR. The occasional -7 on a rogue peak might sound OK, but you better use your ears to confirm this. Adjust the input knob to get in this zone. Then set the output control so that the output meter reads approximately zero dB. You will definitely hear a strong “1176 character” if you set it up this way. It may be too much for your taste, or not enough, depending on what sonic texture you’re going for.

The next item in the signal path, the LA-3A opto compressor, is easier to set up.  With only two knobs, you’d think that everybody would get it right, but this is not always the case. As stated earlier, you have to be careful about driving it too far into gain reduction because the recovery becomes slower—too slow for this series application if the VU meter shows more than one and a half dB of GR. 

On a tangential note, an LA-3A can sound mind-blowingly cool if you slam it with 10 to 20 dB of GR! Of course you will need to crank up the output to make up for lost volume. You’ve heard this sound on hundreds of beloved classic records. It sounds like the guitars on Bryan Adams’ song “Cuts Like a Knife” may have been recorded or mixed with this technique. I may be totally wrong about this being the case, but at least the example gives you an idea of how an LA-3A  can sound larger than life when pushed. 

That, however, is not the way that we are using it in this particular application, in series with an 1176 on a lead vocal. By keeping the LA-3A’s gain reduction at about 1 to 2 dB, we ensure a more organic—yet highly energetic—vocal tone that is the undisputed ruler of the mix. The vocal will still have the illusion of being dynamic, and it won’t require much EQ to make up for the darkness that compression can add.

After experimenting with this technique and tweaking it to your taste, try a variation of it: route LA-3A in parallel instead of series. You can get away with more than twice the amount of gain reduction because you will be blending it to taste (on a separate fader) with the 1176 sound. That means that you can create an aggressive and urgent sound with the 1176, and supplement it with the added sustain and warmth of the LA-3A. Because it’s not in series, you don’t need to worry about it swallowing transients or recovering too slowly.  Just blend it in until you dig the sound, and you’re good to go! 

Another variation on this multi compressor theme is to put an opto compressor first, then an 1176 in series and a second opto in parallel. I’ve had a lot of success with this technique when feeding a Manley ELOP into an 1176 to get a compelling punchy sound, and then blending in a “pushed” LA-3A in parallel to harmonically fatten up the tone. 

Try these techniques and become inspired to create some new ones of your own. Make sure that you experiment with different routings and settings. Remember that my suggested starting settings are just that: suggested, not required. They will be dependent on the program that you intend to process and the mood you want to create, so adjust accordingly and appropriately.

If I were to add just one philosophical thought about compression and limiting in general, it would be that my personal taste is to see the gain reduction meters always “dancing” near zero on the VU meter rather than constantly being driven into deep compression. If you haven’t already done so, experiment with this concept and let me know what you think. 

Whatever you do, in life or in the studio, make sure that you follow your own muse, not somebody else’s. 

These cookie fortunes seem to be strangely related to vocal compression aesthetics.

Multi-Buss Mixing Philosophy 

The question, “What’s on your mix buss?” has been all the rage for several years now. Mix engineers have been pre-mastering their work for a long time, to ensure that the mastering engineer has a clear artistic vision before manipulating the audio.

Digital audio sound quality and mixing “inside the box” (summing in the digital domain) may have caused the latest renewed interest in mix buss processors. Digital sound has often been characterized as harsh and clinical. Character pieces like a vacuum tube compressors and limiters can do a great job of warming up the sound, making it more euphonic. Insert a Manley Labs Variable Mu stereo compressor across the mix buss, and a good mix magically transforms into a juicy record.

That said, stereo mix buss processing has inherent limitations. As an example, let’s pretend that you have set the compressor’s attack and release parameters to make the song have an exciting pumping effect, in time with the music. Think EDM (electronic dance music) as an obvious point of reference. Everything sounds good… until the producer asks you to beef up the low and and add more brightness to the overall mix. All of a sudden your once warm lead vocal starts to sizzle, and the entire mix starts gasping for air every time the kick drum happens. When you are that deeply into a dense mix, adding one more straw can break the camel’s back.

The solution to this problem is actually quite simple. Instead of processing one stereo mix buss, break the mix down into several submixes consisting of components that symbiotically work together. In the video link above, I discuss the workflow of breaking the mix into three stereo buses instead of one. Buss A is for all the vocals. Buss B contains the bass and drums. Buss C includes all the harmonic instruments that are typically panned out to the sides, leaving the center of the soundstage open for maximum vocal, bass, kick and snare punch and clarity.

Each of those three busses is independently processed. They can benefit from using different attack and release settings and thresholds, as well as different EQ curves. 

Perhaps the simplest example of how to deploy this technique would be a scenario in which you want the bass & drums to have an obvious rhythmic “pump” without having the compression affecting the vocals. Further, imagine you want to brighten all the harmonic instruments a lot, without adding sibilance to the vocals. No problem with submix processing!

(Video courtesy of Dangerous Music, Inc.)