Today is Boxing Day which is synonymous with belly ache, boredom and regret. Let us use this day in a constructive and useful way! Either that, or write music. Let’s go:

AAAAA AAAAA AAAAA AAAAA
AAAAA AAAAA AAAAA AAAAA
AAAAA AAAAA AAAAA AAAAA
AAAAA AAAAA AAAAA AAAAA

That’s a really crap song. It just does the same thing over and over again. There’s things we can do to fix it later on, but let’s try again with a little more variety first. Ready?

EMDOE DOANE KFOLE SOQME
CLOAV VWIOF DFGWE LOABD
OQNSJ LEABC DUABD LFKMC
EPVMS AOFBN SLASF SAFQK

That’s no good either. It’s just all random and boring. Let’s think.

You likely have played Peek A Boo with a small child or perhaps a senile great uncle. If you haven’t the rules are very simple.

• Duck out of sight, and pause.
• Suddenly pop into sight.
• Say ‘Peek A Boo’.

Repeating this can keep baby or uncle in paroxysms of joy for hours if done right. It’s all in the timing. The baby knows that you are going to reappear, the mystery is just when. Not too fast – there’s some anxiety and anticipation, but the pay off is certain to come.

‘Classical’ composers are good at Peek A Boo. They give you an overture, which provides the rules. Then they start messing with the melodic structure, wandering off in apparent disregard for the game, but just when you think they’re halfway to China they pop back into view to ensure you know they’re really just teasing. Then they give you a finale that shows how it only seemed to wander off but was really SCIENCE.

In popular music we have the intro, the verse and chorus, the bridge and other structures that provide the same service. But really it’s all about anticipation and reward.

ABCDE ABCDE
ABCAB ABCAB ABCAB CDEFG
ABCAB ABCAB ABCAB CDEFG
MNOPQ CDEFG MNOPQ 12345
MNOPQ CDEFG OPQRS ABCDE
ABCAB ABCAB ABCAB CDEFG
ABCDE ABCDE

ABCDE hello I’m the main riff for the song, watch out for me. Hear me twice to get the idea.
ABCAB I’m part of the main riff but I keep pulling my punch until…
CDEFG peek a boo! I’m the missing bit of the riff and more! Often an extra chord.
MNOPQ I’m the chorus structure, I’m provide variety but I’m usually a variation on the main riff, often shifted in pitch
12345 I’m a special thing that only happens once in the chorus which adds interest. My name’s Wanda.
OPQRS meh, I’m just a way of getting back from the chorus to the main riff.

That’s not a strict map. It’s one of many ways to do the same thing – show the riff, then keep hiding it, shuffling it, then bring it out. The chorus is a way to distract the listener, give the riff a break, which anticipates the next round for more fun. Otherwise you can end up with this:

ABCDE ABCDE FGHIJ KLMNO
ABCDE ABCDE FGHIJ KLMNO
ABCDE ABCDE FGHIJ KLMNO
ABCDE ABCDE FGHIJ KLMNO …

which is basically Don Maclean’s American Pie and one of my least favourite songs of all time. Because it simply plays Peek A Boo at exactly the same pace over and over. You can set your watch by the chorus.

Can we make some simple rules? Kinda sort of. Rules that must be broken and never will be quite right.

Start by establishing the melody in the intro. As soon as you’ve made that clear, start to shuffle it around in the verse, so that it completes only part of the time. The listener should feel like they are watching a card trick, where the Queen of Hearts shows up unexpectedly.

Let’s go for a chorus. It’s usually reminiscent of the verse but it’s off in a different direction. More than likely pitched upwards it swoops up a few times, does a special trick the last time, then it glides back down into the verse structure again. If you do go up, you should usually keep the upper range of the verse structure overlapping the range of the chorus melody so that you can step down through the shared area. Think of a telescopic ladder with two sections that overlap.

You can do that a few times but you might start to get boring. So we…

Break down, drop back to the simple riff again. Stop everything except the main idea.

And then build up.

Here lurks the Gear Change. Now it’s OK near the end to repitch your melody up a third or a fifth to make a nice ‘uplifting’ section that refreshes the music. But like Iraq, it’s all about the exit strategy. Somehow you have to come back down from that high to reconnect with your original riff for the finale. When the writer goes up and never comes down – that’s called the Gear Change. Read this.

If you go up you need to have similar way to get down and likely that used for the chorus. The ladder again.

How to get out at the end? Two thoughts – from Brian Eno – who likes to fade out as if discreetly tiptoeing out of a room where the music continues to play – and from Garry Bradbury – who likes to end it as if an anvil has been hurled at your face.

Now, for our first song. The answer in this case is to do something like this:

aAaAA aAaAA
aaaaa AAAAA aaaaa AAAA
aaaaa AAAAA aaaaa AAAA

Using tonal adjustments to create the progression instead of pitch. But unpitched composition is wide field and we will have to come back to this later. Meanwhile, more ham.

Vocals are, as you have become used to me saying, made up of multiple sounds. Look at a head.

Here are a set of cavities. The throat, the mouth, the nasal cavity etc. Each has a size which resonates at a certain pitch and they are joined up in sequence. If you are Barry White your big cavities are resonating low, and if you are me, not so much. As Barry sings, he creates a fixed note cluster that colours the notes he sings, which is the formant. If you are Warren Burt or Diamanda Galas you have learned how to produce two notes at once by independently resonating your nasal cavity. Others may have perfected borbarigmus, I hide their secret here.

The tone of your voice stays much the same no matter what note you sing (although different volumes and stresses accent different cavities). Pitch shifting the voice up and down unnaturally shifts this tone such that you get the familiar ‘giant’ and ‘chipmunk’ side effects. Modern samplers and Pro Tool’s pitch shifter offer a formant adjustment that compensates for this, keeping the cavity sizes constant when shifting the pitch. Also you may impose an artificial formant on a voice to create a larger or smaller vocal apparatus – the basis of voice changer software.

Recording in a room captures not only the sounds you’re after but the tone of the room, the resonance of the room itself. Which is is why when you come back and redub dialogue into a film work it may be  unconvincing – just as reshooting under a different temperature of light will look wrong, recording in a studio won’t match up with the live sound. You need to always capture a little ‘air’ from your location to cover the rerecorded sections and very likely will have to EQ the ADR voice to try match the tone of the room. Very often once you’ve identified the room’s pitch you can create a peak (increased volume closely around a certain frequency) in the EQ that is similar – although notches (decreased volume) have sometimes worked better for me, and often as comb filters – it depends on the nodes in the room – more later.

Musicians should be aware of the resonance of the room when finding the best place to record an instrument – singing in the shower – or like Joy Division, recording the drums on the roof. Flat acoustics are not always the best.

3D animators (only those poor souls that have to do their own sound design) need to think of the rendering of sound as much as the vision. I ask my 3D kunstlers to use a fixed flanger (which is a kind of comb) on all the foley in their space epics to create the metal walls and floor we’re seeing on screen. Done right, the reality is much more powerful.

There are holes in the head for the gas to get out – cake hole and two nose holes (ears don’t count). They add more upper ‘air’ to the sound as the breath collides with tounge and teeth and lips. One is able to sing more through the nose or mouth. You can even sing with both closed. Try blocking each and listen to the ‘submixes’ of your voice. (Please remember to unblock nose and mouth when finished).

When we sing louder the formant shifts slightly and we add more or less noise into our mix. One popular trick especially with female singers is to add a little white noise into their vocals, expanded by the amplitude of the voice. This breathy voice sounds more passionate and intimate – as if close mic’ed but without the bass shift that brings. Compressing the voice can also be used to apparently shift the emphasis and was a major part of being a Beatle.

Recording dialogue in a room is therefore quite tricky – resonances everywhere, microphone placements causing filtration etc. It’s suprising how often people get decent results, but they do. Getting beyond that takes years of listening and experimentation. As an engineer you have to think like a director of photography, but instead of seeing light and layout, you have to hear the tone of the space, the sounds from outside the space and the way the voices reflect off the surfaces. And more.

Go to the space, close your eyes. Listen.

You play a note on a bass. (Here we’ll only look at the note and not any of the higher tones that add to the timbre of the instrument.) The bass string vibrates back and forth, it pushes together air in a radiating pattern of compressed and uncompressed regions that reach our ears and cause the eardrum to vibrate in time. Drawing pockets of air is difficult so we draw a waveform that shows pressure over time.

All is simple so long as that bass note reaches our ears undisturbed. But suppose we had two people playing exactly the same note on the bass. They are slightly different distances from our ear.

Depending on the distance to the source the two waves arrive more or less out of phase. The more out of phase they are, the less volume we get as they add together, as the peaks and troughs cancel out each other. So what chance that two basses play exactly the same note? Very little. However the same bass recorded on two mics or coming out of stereo speakers… entirely likely.

Areas where the signals cancel out we call a node. You will find multiple nodes in a stereo listening space where bass is going to disappear or be muted. If you are micing up a kick or bass you need to avoid placing mics that cancel out some or all of the sound. Monitoring the combined mic inputs in mono can help detect phase problems. But better use just one microphone. Having bass coming out over stereo speakers is going to create nodes as well and make your mix sound bad as people move around – big problem with live shows. But we can try create a large ‘sweet spot’ by limiting the stereo spread of low frequencies. Make your kick come out the centre of the stereo space, so that both speakers are moving as one. Basically treat low frequencies as a mono recording and avoid stereo treatments.

Why only bass? Because as you go up in frequency these nodes get so close together that you are less likely to notice them. Also the richer the sound (in terms of multiple overtones and sounds in the instrument) the more complex the pattern of interference becomes. That can be an advantage if we try to use higher pitched sounds within the instrument.

Because we usually place the bass frequencies in the middle (unless you are going for a particular effect of course), the instruments start their arrangement out from a central ‘shaft’. This diagram is going to be refined greatly as we go along, but you can already see how the bass instruments push the higher pitched ones left and right in a tree shape.

In the early days of stereo people would mix all the bass guitar in one speaker. That works as well. But if you are going to have it coming from two places then better to have two different sounds – e.g. the deeper part of the sound here and the slap over there or use effects such as phasing that introduce radical changes in the phase of the instrument. Recording the same part twice is good too.

And this is yet another reason why mixing on headphones can be dangerous … no phase cancellation between the left and right phone. Use speakers to check for this.

Much as I love to laugh at people all day, let’s use this blog for something useful.

Most people have a general idea about frequencies of sounds. They know that some sounds are ‘low’, ‘deep’ and some are ‘mid’ and some are ‘high pitched’. When they come to mix music they know that different instruments can be used together when they don’t overlap too much in frequency. They know an orchestra is designed so that different instrument groups such as horns and strings don’t compete with each other. That’s absolutely right, but it’s just the starting point of understanding how a music mix works. We can improve our music recordings greatly by a finer understanding of frequency.

When mixing I like to use a diagram. It has the Y axis, the one that points up the page as the frequency. The X axis, the horizontal one, is the panning across the stereo field. Implied here is a Z axis, which would extend ‘into’ the paper, of time. Our diagram is a slice through a changing pattern across the X and Y axes. We’ll not worry about the Z, but might have to draw more than one diagram per song.

If the sound is ‘low’ in frequency, it goes at the bottom of the diagram. A kick drum for example, goes roughly at the bottom. Hi hats go up the top. The way frequency works is such that as we go up the Y axis the frequency scale ‘speeds up’ logarithmically. There’s more change at the top than the bottom. Not too complex so far.

Each sound has a fundamental note, the basic tone of the sound. Above it come a set of overtones that give the sound a timbre. The tone of an electric piano has (ignoring some detail) few overtones above the fundamental, it’s ’round’ or ‘soft’. The scrape of a violin has many more overtones and is ‘sharper’. A gong has many overtones which are inharmonious with the fundamental tone. A cymbal, which is a type of gong, has so many overtones it makes a sizzling sound.

So in our diagram, we can’t place the instruments in one place, rather they cover an area of the frequency range. And it’s those areas that can overlap. If your kick drum has overtones that extend into the region taken up by a bass guitar, they may be masked by the bass, making the kick sound less rich. This is a reason why a mix can sound muddy. But there’s more.

While synthetic instruments may have one fundamental sound, very rarely do real instruments occupy just one area of the frequency spectrum. They have at least two and often more sounds that combine to give them their identity. For the purposes of our discussion here, let’s use the term ‘sound’ for one component in an ‘instrument’.

Draw a picture of a snare drum. There’s the skin at the top. There’s the drum. There’s the snares under the drum that rattle. That’s three sounds in one instrument. When we place that snare in our mix diagram it’s going to take up multiple spots. The drum will ring somewhere around the middle, an area that we can roughly identify as the ‘vocal’ range. That ring is a tone with mostly harmonic overtones. The whack of the stick on the skin is somewhat higher up. Even higher is the rattle of the snares. All three are now taking up space. When a recording engineer uses a microphone above or below (or both) a snare, she’s trying to accent one of these multiple sounds so that the mix can choose between them.

A kick drum has a skin and a drum but no snares. It has the boom of the drum which is a very deep note, also the thwack of the pedal hitting the skin. All instruments use up multiple areas of the frequency domain. So when somebody says an instrument is ‘at’ a certain frequency they’re simplifying.

EQ works in accenting or reducing overtones in the sounds that make up an instrument. Rolling off the top end of a kick drum removes the pedal sound, then as we go down, the drum becomes less sharp. Rolling off the bottom end we lose the boom of the drum first and eventually hear only the impact of the pedal. We can also ‘notch’ a frequency to remove a particular sound from an instrument.

So how does knowing this help mixing? For now I’ll do an executive summary, then later we can go into the specifics.

We can write music that doesn’t try to have similar instruments playing at once. e.g. Write bass notes that don’t sit on the kick, but oppose it or anticipate it.

We can also write it so that more instruments join in later and our memory of what went before gives the impression that we still hear the earlier instruments clearly. The sound can drop out to remind the listener of what was concealed.

We can move instruments out of each others way by either choosing them wisely (bass, guitar, drums, but no flugelhorn), recording them to occupy different frequency spaces, or using EQ to do the same thing. The reason that rock bands tend to play a fixed set of instruments reflects years of experience. Also a guitarist knows to switch models so as to shift his timbre out of the way of other instruments. As a recording engineer you should be listening to sounds for their timbres and trying out EQ to change these in a reproducible way. We’ll talk about EQ more later.

More advanced, we can remove a specific area of frequency from one instrument that was masked by a another. When two sounds try to occupy the one space, they might add together, but they can also subtract from each other if they are out of phase, again explained later. Usually we get ever changing phasing that causes that frequency to oscillate unpleasantly. When we remove that frequency from an instrument (by for example not using the ring of a snare) we get a cleaner sound. The listener doesn’t notice the missing frequency because they hear it covered by other instrument.

We can move sounds into different areas of the stereo field. Drums in the middle, guitars at sides. There’s all kinds of rules here that we will come back to. Let’s just say that the ears are assisted greatly by having sounds come from different origins. We’re built to discern from where a sound comes, and so the frequencies untangle as they move across the X axis.

We can accent parts of the instrument, for example place a microphone so that the slap of a bass is louder than the bass body, or use compression with an attack so that the slap is made louder, and more on compression later.

We can add overtones to a sound by using overdrives. That’s what a lead guitar does to stand out. When ‘industrial’ bands use distortion on every instrument, they wonder why it ends up sounding dull – simply because again everything is overlapping.

We have quite a few things to cover just on this idea and for that reason we’ll meet again in the next installment. Anything tagged in the blog as ‘music class’ will be useful. Anything else you can be sure is rubbish.