Shaker!

Shaker chain: Audio Technica 4050 > Neve 1073 > Pro Tools

01/20/12

Mixing Philosophy (Part 2): Stereo Width

In part 1, I talked a lot about the different frequency ranges and what sorts of tones and timbres reside in them. In part 2, I’ll be getting deeper in to different forms of processing that could be applied to those ranges in order to accomplish a very specific goal: mixing with dimension, specifically with three dimensions.

What does mixing with dimension mean? My take on it is that a mix which is well defined in all three dimensions will have certain desirable characteristics that a less well dimensioned mix will lack. Among the things that I perceive as desirable are: good amounts of space in the mix, proper separation of instruments, movement, dynamics, and presence. All of these components contribute to a general feeling that the mix is no longer “flat”, but rather it is able to sort of come out of the speakers and envelop the listener inside of the mix and provide a suitable soundstage for all of the instruments and mix elements to be heard and enjoyed.

So, what are these three dimensions? They’ve got lots of names, but the easiest way that I’ve found to conceptualize them is to think of them in terms of how we experience our three dimensions of reality in life: Width, Height, and Depth. 

Width is the distance from one side of something to the other. In mixing, we create width through panning and phase interaction. In short, this if movement/spacing from left to right.

Height is the spacing from bottom to top in a mix. We can achieve this through equalization. If you think about it, we’re all accustomed to calling the bass “low” and the treble frequencies “high”, so we’re already set up to think about sound this way, but many people have never considered frequency as a way to create dimension in a mix.

Last but not least is depth. This is an easy one to describe: it’s the distance from front to back in a mix, or the way some things sound like they’re farther away from the listener than others. Depth is created through the careful use of time based effects like reverb and delay and to some extent is also effected by equalization and volume.   

In this post, I’ll be talking specifically about width. I was was going to write about all three dimensions…but that’s going to be way too much to read in one sitting (not to mention too much for me to type). So let’s get started with width. We’re all familiar with the pan knob in our DAW or on our mixing board. They all basically do the same thing in that they move a sound across the stereo field from one speaker to the other. It is perhaps the most basic element of mix dimension to understand, but it can be difficult to really control. 

First let’s talk about where panning came from. Originally, sound was in mono (only coming from one channel our source) there was no left or right side, there was just a speaker (or a couple speakers, but all playing the same thing). As you can imagine, this made for a less than ideal listening experience, but due to technical limitations, that was all that was available. Some sources such as AM radio are still in mono. So when was stereo invented? Surprisingly, it was way back in 1881! Ok…it wasn’t actually used in a recording context, but a guy named Clement Ader used a system of telephone transmitters in the Paris Opera house to connect two microphones on stage to a suite of rooms at the Paris Electrical Exhibition where listeners could hear a live transmission of the performance through a pair of receivers that were placed on each ear. Ok, yeah I know that doesn’t really have to do with recording, but now you can impress all your friends with your extensive (if slightly nerdy) knowledge of stereo history, so that counts for something, right?? Moving on to recording, stereophonic records didn’t really come on the scene until the mid 1950’s and became fairly standard in 1957 and by 1968 most of the major record labels had stopped producing mono recordings. However, even as stereo became prevalent, no one necessarily knew what to do with it, and to make things even more awkward, early stereo recording consoles didn’t have a pan knob like we have now, they came equipped with a pan switch. Yes, a switch. They typically had three choices: left, right, or center (both speakers) and that was it. That’s why on a lot of early Beatles recordings you’ll hear the drums coming completely from one side, and maybe guitar or vocal completely from the other speaker. Moving in to the modern era, we’ve got pan knobs! This means we have the freedom to place our mix elements in either speaker or anywhere in between. In creating a 3D mix, this is very important in order to avoid the perception that the mix is identical from side to side which would lead to a mono sound. Don’t be afraid to pan things to the extreme in order to create space. I’ll often pan mix elements very hard to create separation and if I’ve doubled something (like a guitar part) I’ll usually pan one take hard left and the other hard right to create a larger sense of space. Please note, when I say I’ve doubled something, I don’t mean I’ve duplicated the track, I mean I’ve actually had the guitarist play the same part two times and panned them to opposite sides. If I were to take a track and simply duplicate it and pan it, I end up with what I refer to as “big mono”. The part sounds louder, but it doesn’t actually get wider because the part is identical on both sides. What creates a sense of space is the slight variations in pitch an timing that will invariably happen wen a player plays a part twice. It’s those differences that let the ear know that there is variation, which our brains perceive as width. 

Speaking of our brains, there are two psychoacoustic (how our brain interprets sound) phenomenon that we as mixers can take advantage of. The first is called the Haas Effect, and it was described in a 1949 Ph.D thesis by Dr. Helmut Haas. Besides being automatically cool by simply having the first name Helmut, Dr. Haas was cool because he basically discovered that humans localize the source of a sound (aka decide which direction a sound came from) from the first sound they hear, even if another sound comes from a completely different direction shortly thereafter. 

Wait…what? That sounded like science or something….

OK, here’s an example: you’ve got a speaker on each side of you, a single sound is played through both speakers, however, one side is slightly delayed. Even though the sound is the same and is coming from both sides, your brain will still say the sound came from the side that you heard first (the one that wasn’t delayed).

Here’s where it gets really weird, that second, delayed sound can actually be up to 10dB louder than the non-delayed sound….and your brain will STILL say that the sound came from the side it heard first, even though the slightly delayed side was louder. 

Pretty weird right? So why does it happen? The answer lies in a very simple fact: people have two ears. In life, when sound comes from the left, it reaches and is processed by the left ear slightly before the right ear because it takes sound a little while to travel around your head. Due to this delay, the brain interprets this as “a sound coming from the left”. We can use this effect to our advantage to created space in the width or our mixes. For example, we can easily set up a send to a delay plugin on an aux track that will delay the sound by a small amount like 10ms. We’ll pan the original track to the left, and the delayed track to the right….and boom: Haas effect in action. The sound will appear to come from the left, even though it’s equally loud on both sides. This can make sources that are mono, seem to be more stereo even though it’s actually an illusion. 

Haas found out something else that’s pretty cool too: our brains are able to localize a sound even more effectively if frequency is involved. In short, our brain think that a sound is more directional based on how much treble there is. Once again, this is because we have two ears. In real life, when a sound comes from the left, it not only hits the left ear first like we already talked about, but it also loses high frequencies as it has to travel around our head from the left to the right. We can replicate this in mixing. Take the example from earlier where we had an identical signal duplicated and panned hard left and right but with the right side delayed by 10ms. We already know our brain will localize this signal as coming from the left, but now, take an eq and cut some of the high frequencies out of the delayed right side and our brain will be even more convinced that the sound is coming from the left! It’s really pretty amazing, and utilizing this technique can give your mixes a huge sense of having more width naturally in a way that simple left to right panning can’t replicate.

Ok, you still with me? I’ve got one last thing: this is the second psychoacoustic process that I mentioned earlier and I’m going to broadly describe it as “Mid-Side Techniques” or M/S Techniques. You’ll see these all over the place in plugins marketed as stereo wideners like the Waves S1, the Brainworx BX_Solo and many others. They don’t all do exactly the same thing, but often it’s a variation on what I’m about to describe. Basically a mid-side processor is capable of separating a stereo signal in to (technically) three signals. The first is the Mid (aka middle) signal is the information that is found in both sides of the original stereo track (like the vocal, kick, snare, bass…anything that’s panned right up the middle) the side channels are created by taking the newly created Mid channel and either adding or subtracting the common data from the original left or right track. Basically the math looks like this if your really curious:

L=(m-s)

R=(m+s)

There’s a really good description about it right here on the Brainworx site for those who are interested:

http://www.brainworx-music.de/en/whatisms

It’s got a fancy little diagram and everything!

In essence, this sort of processing allows you to modify the sides of a signal without altering what is in the middle or the opposite (modifying the middle without messing with the sides). In mastering a common function might be if a singer has an “Ess” problem and you want to put a de-esser on him but all you have is a stereo file of the whole mix, you can run it through a mid side processor and put a de-esser just on the mid signal (where the lead vocal usually is) without effecting any of the other elements of the mix. 

Now, what does this have to do with our topic of width? Actually a lot. Once a signal has been M/S processed, we can raise the level of the sides and create the appearance of a wider mix without altering the levels very much. I’ll often put this processing on a reverb return to make a sound source appear to be in a hugely wide room or to make certain elements like synths and pianos actually seem to stretch outside of the speakers for and ultra wide soundstage. However, this does come with a couple warnings: first, too much of this sort of processing and peoples ears will start to pick up that something isn’t quite right, that sounds are too far apart to be real and it will ruin the illusion. Second, since much of this processing is achieved through inverting the phase on one side of the signal, too much of this processing can make it where the sound is no longer mono compatible. This may not be a huge issue depending on what your listeners will be listening to, but if you’re ever doing a radio spot for AM radio you have to watch out. I once heard a car commercial that was on FM (stereo) radio and featured an announcer describing all kinds of deals and then a massive cheesy explosion at the end…I assume that was to punctuate that the deals were explosively good or something like that, but anyway, it sounded cool in stereo, the announcer’s voice sounded MASSIVELY wide…..then I heard it on AM radio. Or rather, I didn’t hear it. The announcer’s voice had been made to sound huge through phase based widening and had ended up completely canceling itself out in mono so all you could hear was about 20 seconds of indistinct, phase-y mumbling and then…..a huge explosion….which was apparently the only thing in the whole commercial that hadn’t been widened like crazy. It was like 20 seconds of phase-y embarrassment punctuated by two seconds of explosion based ridicule. I guarantee whoever did that commercial didn’t get called for another one. 

Alright kids, that’s width! Coming next is height! Get ready for it!

01/11/12

Mixing Philosophy (Part 1)

Here’s a concise (hopefully) exploration of what I listen for while I’m mixing. I say exploration because I’ve realized there are a lot of things that I do in my mixes just because I like the way they sound, but I’ve never really thought about why, so my goal in this is to be able to explain not only what I do, but when/why I do it so that it can apply to your mixes as well. 

And…I just added “Part 1” to the title. I’m not even through with the introduction and I’m already deciding there’s going to be a part 2. I lied, this won’t be concise. Not at all.

First, different frequency ranges:

Music is just a bunch of frequencies coming together in a way that’s pleasing to our ear. So what makes some frequencies desirable and others not? Why is there such a thing as an equalizer? Let’s begin our exploration shall we?

I break frequencies down in to five ranges, you’ve probably heard them before: Low, Low-Mid, Mid, High-Mid, and High frequencies. For the sake of this article, lets define those ranges: Bass will be 20hz to 200hz, Low Mid are 200hz to 600hz, Mids are from 600hz to 3kHz, High Mids are 3k to 7k and Highs are 7k to 20k.

You’ll notice there are a lot of mids in there. That’s not by accident. The midrange is where human hearing is most sensitive and accurate and is most able to tell changes in frequency. Therefor, we have lots of names to describe this range.

Let’s start with the low end. This is the foundation of your mix. It’s the “oomf” that shakes the car and makes drunk people dance poorly. It’s also one of the trickier ranges to control because there’s a lot of sound energy in this range. Since sound is vibration, you’ve probably guessed by now that the low end can create very powerful vibrations (like what shakes your car) and while that power is important to make your mix sound full, thick, and heavy, it can also easily overpower a mix if it’s allowed to run wild. For this reason, I tend to insert a high pass filter on the majority of my tracks that I suspect contain extraneous bass information. Things like acoustic and electric guitars, vocals, snare drums, and basically anything that isn’t a “bass instrument” like a kick drum, bass guitar, etc. I do this to prevent the buildup of unnecessary bass frequencies that could contribute to the low end overpowering a mix. I think the most important word to remember when listening to the bass is “balance”. The bass, as booming as you might want it, must be proportionate to the other frequencies. It can still boom, but it can’t overpower. 

Next, Low Mids. This is one of my favorite frequency ranges to talk about because so many people go around scooping out low mids because somebody told them that 250hz was designed by the devil himself trying to make their mix muddy…..and then they proceed to drain all the warmth out of their mix. Yes, the low mids can be muddy. They can also be warm. The word I think about for low mids is “control”. Control in the level, control in the dynamics, and control in the tone and resonance. Additionally, the low mids are what tend to balance out the other sections midrange and like we talked about before, the midrange is the area our ears are most sensitive to, so without a balanced amount of low mids, other frequencies can start to feel overly abrasive when you turn the mix up.

Third, Mids. This section of the midrange is where a lot of the fundamental clarity of your melodic instruments come from. It’s also a range with many distinct tones within it. For example, the “honk” of certain instruments comes in around 600hz-800hz, paper-y sounds come in around 800hz-1k, 1k to 2.5k has got what I affectionately refer to as the “walkie talkie” frequency. Along side that and extending slightly above is the “tinny” sound that some instruments can have. All of those descriptions sound rather negative, and they are, but that’s because the human ear tends to notice a given frequency when it’s too loud to where it becomes abrasive (and therefor painfully obvious) so we’ve adopted these descriptors, but conversely, a lack of these frequencies won’t sound right either. As I said earlier, it’s where a lot of the clarity and definition are and for that reason they are vitally important to a good mix. I thought for quite a while on what word I’d associate with the midrange, but after thinking about it, I’m going to go with “flavor”. To me, this has two meanings: first, it describes the variety of timbres that are available in this range, second, it reminds us that even within this particular frequency range, one flavor, or frequency, can easily overpower the others and a midrange that is in poor taste. (See what I did there? Punny, right?) 

Next are the High Mids. I’m going to go ahead and give you the word association right here because it’s pretty obvious: moderation. This is the frequency range that the human ear is most sensitive to, and thus, is a range that can easily fatigue the ear if there’s too much of it. Now, go Google “Fletcher Munson Curve”…yeah, right now. It’s really important. I’ll wait, I promise. Ok, did you see that? It’s a graph explaining how much sound energy it takes for humans to perceive frequencies at “equal loudness”. What the heck does that mean? Well, it basically answers the question “If I have a sine wave at 100hz coming out of one speaker, and a sine wave at 2.5khz coming out of another speaker, how loud would I have to turn one of those signals to make it sound the same volume as the other?” Somewhat surprisingly (or perhaps not if you Googled what I told you to) it’s not going to be the same volume for both signals. You’ll actually have to turn the bass signal up significantly more than the high signal to make it sound the same. Why is this? It’s because our ears perceive some frequencies more easily than others. Specifically, it takes a lot of low end for us to be able to hear it, then progressively, we hear the frequencies more and more easily until finally we reach our most sensitive spot: the upper midrange. Scientifically, this is because our ear canal resonates at around 2.5k to 6k and sort of acts as a natural amplifier for those frequencies. See, science can make you mix better! Isn’t that cool? Armed with this newfound information, you can now understand why we are most sensitive to this range and why it leads to fatigue and why it must be used in moderation. If you solo electric guitar, or a voice or, a snare drum and start to eq it, adding a good bit of these upper mids may sound pretty appealing. It causes the element to come forward in the mix, there’s more presence, and it just sounds….better. Then…you un-solo everything. And your mix rips your head off and you die in agony and the last sound you hear is the abrasive harshness searing your eardrums. Ok…I exaggerate. Point being, this is actually a frequency range that can sound very good and is vitally important in creating a mix with good presence and punch, but it’s also probably the easiest frequency range to overdo, and the buildup of frequencies across this range will creep up on you until you’re left with a harsh, fizzy, and abrasive mix. Use it in moderation and you can bring elements forward and add presence, abuse it and you’ll hurt people. It’s that simple, and that complicated.

Alright, last one: Highs. This range is tricky; it’s not quite as dangerous on the ears as the high mids, but there’s still an element of risk if this range is overused. Too much and you’ll run in to tasty things like sibilance (around 6-8k) and brittleness (around 8-16k) above that, it’s almost frequencies that are felt rather than heard. The easiest way to figure out what those high frequencies are like is to insert a low pass filter and cut some of them out and note which sounds and timbres go away. Breathiness in vocals is located here, as is the snap of the snare. Much like the low frequencies, these can build up across a mix, so a lot of times I’ll choose to put a gentle low pass filter across some of the elements that just don’t need to be up that high (especially in close mic’d drums to eliminate some of the tones of the cymbals that may bleed in). My word association for this range is “Contrast”. In this syntax, contrast refers to the difference the high end can make in whether the sound appears bight or dark (aka dull) and also to how useful this range is in creating differences between elements in the mix. Some elements are lighter, some are darker, but they all create an overall sense of depth, space, and movement in a mix.

Alright, that wasn’t concise at all. I’m sure part two will be even longer. In that, I’ll talk about how a mix can come together across three dimensions: High to Low (achieved through different frequencies) Left to Right (achieved through panning and phase) and Front To Back (Achieved via volume, effects, and frequency)

Until then, here’s a recap on the word associations:

Low = Balance

Low Mid = Control

Mid = Flavor

High Mid = Moderation

High = Contrast

Thanks for reading all this!

-Johnny

01/10/12