Arrangement

By the end of chapter 19 you had everything a song is made of: a tune, chords under it and a bass, and since chapter 9 you have had a beat. Play them all together and you find that four good parts do not yet make a track. They jostle and shout over each other, and after a minute the same loop wears thin. Turning a set of parts into a track is the job of arrangement: deciding who plays what, with which sound, how high and when.

This chapter is one day in a studio. In the morning there is a project with everything in it that nobody can bear to listen to. We take it apart and build it again, an hour for each track, and each track gets its own job and its own sound from the synth of chapter 2. By the evening the four-bar loop will have grown into a track with an intro, verses, a chorus and an outro. Put on headphones: half of what follows lives below 200 Hz, and a phone speaker barely reaches down there.

9:00. All there, and unlistenable

Here is the project the day starts with. Four bars, the chords A minor, F, C and G, a tempo of 100 BPM. Press Play and listen to a couple of rounds.

These are the same notes you will hear in the evening, but they are hard work to listen to. The bass sounds somewhere in the middle, the melody has sunk into the chords, and everything blurs into one drone. Under the buttons is the spectrum of each track in its own colour, and it shows why: all the lines have crowded into one band, from 200 to 600 Hz. Red hatching marks the frequencies where two tracks are equally loud. There the ear cannot tell them apart.

Now switch on "Separated". Not one note has changed. The bass has dropped an octave and got a proper bass sound, the melody has moved up an octave, and the chords have gone to a soft pad. The spectrum has spread out into floors, and you can hear each part on its own. Try the M buttons (mute the track) and the S buttons (solo it, so only that track plays). That is how you check who plays what on any mixing desk and in any DAW.

Four roles

Before building it again, share out the work. In almost any track, from a pop song to techno, the parts fall into four roles. You can hear them in the 9:00 project: switch on "Separated" and press S on each track in turn.

• Rhythm keeps time: kick, snare, hi-hat. It is what people dance to. It is covered in chapter 6 and chapter 9.
• Bass stands on the bottom floor and ties the rhythm to the harmony: it plays the root of the chord, and in the rhythm of the kick. This is the bass line of chapters 12 and 19.
• Harmony is the chords in the middle. It sets the mood and the colour and rarely draws attention to itself: it is the background the melody stands out against.
• Melody sits on top. It is what the listener takes away and recognises from the first few notes.

A role is a job in the track, not an instrument. A guitar can be harmony or melody, and a synth can be anything. It works best when one part does each job and stays out of the others' way. Two parts doing the same thing at the same pitch do not sound twice as good. They sound twice as crowded.

10:00. Rhythm

A new project, empty. The rhythm track goes in first, because everything else will lean on it. Choose a beat. "Light" has the kick on beats one and three and a rimshot on two and four. "Rock" is the backbeat you already know. "House" is four on the floor, a kick on every beat.

Look at the spectrum: the drums take up both edges. The kick is a short sine wave that falls from 150 to 45 Hz in about a tenth of a second, the very bottom. The hi-hat is noise above 7 kHz, the very top. The snare and the clap are in the middle, but only in short bursts. For most of the time the middle stays free, and that is good: the chords and the melody will live there. Switch between the beats and find the kick and the hi-hat on the spectrum. The course will remember the beat you choose until the evening.

11:00. Bass

The second track is the bass. It plays the root of each chord: A, F, C, G. Its rhythm is not random. The bass notes fall where the kick does, on "one", on "three" and on the eighth before beat four, with a short fifth at the end of the bar. Together the bass and the kick give the bounce you put together in the bass builder in chapter 19.

Try the sounds one after another. The "bass" sound from chapter 2 is a sawtooth with a square an octave below, a half-closed filter and a short filter envelope that puts a click at the start of each note. It is full at the bottom and hardly reaches up at all. A "pad" in the bass's place swells in too slowly and drags into the next note, so the bass turns into a drone. The "lead" is bright and thin, and the "keys" die away before the note ends. Set the octave to "+1" and compare: the bass has moved up, and at once it feels as if someone has taken the floor out of the track.

The stretch of pitch a part plays in is called its register. The bass has a low register: the fundamentals of its notes here run from 65 to 165 Hz, and the "bass" sound adds a second oscillator an octave below. The kick already lives on this floor, which is exactly why the bass and the kick are put in the same rhythm. If they have to share the bottom anyway, they should at least hit together rather than take turns.

12:00. Harmony

The third track is the chords. Here they sound as whole notes, one chord to a bar, voiced so that the fingers move as little as possible: A–C–E, A–C–F, G–C–E, G–B–D. Neighbouring chords share notes, and the harmony flows smoothly, without jumps. This is the voice leading of chapter 15. While harmony like this is playing you hardly notice it. Press M on the harmony and you notice at once.

A sound for chords that sit underneath everything is called a pad, like a cushion. A slow attack of almost half a second, a long release, two slightly detuned oscillators and a filter that breathes quietly under an LFO. A pad does not start with a hit, it fades in, so it does not compete with the rhythm. Compare it with "keys": the chords turn into hits and compete with the drums for attention.

Play this harmony yourself: four chords round and round. Use any inversion that suits your hand. The keyboard will show you the notes.

13:00. Melody

The last of the four tracks is the melody. It is built from chord tones with passing notes between them, and it steps down in every bar: E–D–C, F–E–C, G–E–C. The fourth bar ends on G, so that the loop wants to start again. How to write motifs like these was the subject of chapter 18.

The synth sound for a melody is called a lead, because it leads. It has to be heard over everything else: a fast attack, a high cutoff, a little vibrato from the LFO, a touch of glide between notes. Set the octave to "−1". The melody has dropped onto the pad's floor, and the pad starts to swallow it: the level is the same, but it is harder to make out. Give the melody the "pad" sound and it smears and lags behind the beat.

All four tracks are in. The 13:00 project sounds just like "Separated" did at 9:00, only now you know why each track is the way it is.

14:00. Who is in whose way

Why does the melody get lost an octave lower if its level has not changed? Because every note has overtones above its fundamental: at twice the frequency, three times, four times. A pad chord at 200–350 Hz drags overtones with it up to several kilohertz, and they land right where the melody plays. When one part is louder than another in some frequency band, the ear hears the loud one and stops picking out the quiet one. This is masking: the loud sound hides the quiet one, the way a street lamp hides the stars.

The map below works this out without sound. Across is frequency, with a keyboard along the bottom: on this scale every octave is the same width, and the keys are evenly spaced. Each coloured curve shows where a part's energy is, calculated from the waveforms of its sound and from its filter. The bar under the curves is where the notes themselves are. The dotted line is the drums.

Press "Mud": the three bars of notes pile into one, and the verdicts under the map turn red. The margin is how many decibels louder a part is, on average, than its neighbours in the band of its own notes. Go back to "Separated" and move the octaves with the "−" and "+" buttons. Each step shifts a bar by one octave's width, and the verdicts are worked out again.

Leave the octaves where they are and bring the pad's cutoff down. Its curve sags on the right, the upper overtones go, and the melody's margin grows by a few decibels, though not a single note has moved. This is the same low-pass filter from chapter 2, only now it is working on how crowded the track is rather than on the sound. The "Dark pad" button sets the cutoff to 500 Hz. Press Play and compare by ear: the melody comes forward, though nobody has touched it.

The melody is drowning behind the pad, even though it is louder. What is most likely to help?

Bring the pad's filter cutoff down Move the pad up an octave Add resonance to the pad Give the melody a slow attack

What hides the melody is the pad's upper overtones, and a low-pass filter removes exactly those. A pad moved up an octave lands on the melody's floor and gets in the way even more. Resonance boosts the overtones around the cutoff, which makes the pad brighter. A slow attack blurs the start of the melody's notes and makes it harder to hear.

Now by ear. One track from the song plays, each time in a new key and a new variation. Decide its role: rhythm, bass, harmony or melody. The clues are the register, the length of the notes and whether the part moves.

15:00. A sound for the role

You can choose a synth sound, a patch, for a role almost by calculation rather than by taste. A role has a typical note length and a typical pitch. A sound has an attack, a release and a cutoff. If the attack is longer than the note, the sound never reaches full level. If the release is longer than the gap between notes, the notes run together. If the cutoff lets twenty overtones of the bass through, the bass climbs onto the chords' floor.

In the table below the rows are roles and the columns are four synth sounds. A tick marks the cells where the sound passes all these checks. Tap any cell: that part will play with that sound inside the track, and under the table you get a breakdown with figures from the patch itself.

Press "All swapped": the bass gets the pad, the chords get the pluck, the melody gets the bass, and the running arpeggio part gets the lead. The notes are the same, and the track falls apart: the bass drones, the chords die a third of a second into each bar, the melody sounds muffled. The fourth sound in the table is the pluck: a hit, a bright start from the filter envelope and a fall to silence in a third of a second. It is no good for long chords, but for fast repeated notes it is the best sound there is. We will come back to it in an hour.

The four role sounds are easy to recognise even without a part, from a single note. Listen: which sound is it?

And now the other way round: the sounds are mixed up on purpose. The bass may play as a lead, the melody as a bass, the chords as a bell. You can still hear the role if you listen to the register and the way the part moves rather than to the sound.

One track plays, but with the wrong sound: the bass may sound like a lead, the melody like a bass. Listen to the register and the movement, not the timbre. Which role is it?

You can also train your ear for timbre on its own in the "Timbre by ear" trainer.

16:00. The arpeggiator

A pad holds the whole chord. But a chord can also be broken up: its notes played one at a time, quickly and round and round. That is how a harp plays, hence the name: a broken chord is called an arpeggio (from the Italian arpa, harp). Chords themselves are the subject of chapter 15. A synth can break up a chord by itself: you hold the keys, and it plays their notes one after another, exactly in tempo. That is an arpeggiator.

Tap a chord button, or switch the arpeggiator on and hold a chord on the keyboard. With "latch" on, the chord keeps going after you let go of the keys, so you can play with one hand and turn knobs with the other. The next chord you press after letting go replaces it.

Go through the controls and listen to what each one changes. Pattern: up, down, up-down, random, or in the order you pressed the keys. Rate: quarters, eighths, eighth-note triplets or sixteenths; at 110 BPM a sixteenth lasts 136 ms. Octaves: the pattern repeats one or two octaves higher before it comes back. Gate, the length of each note: at 10 % the notes are short and spiky, at 100 % they join up. Play the four chords of the track, a bar each, with the beat: that is already a finished part, though your fingers are doing exactly what they did in the harmony exercise.

The pattern is up-down, two octaves, and you are holding C–E–G. How many notes sound before the pattern starts again?

10 6 12 11

Up through two octaves is six notes: C, E, G, and the same three an octave higher. Back down is four: the pattern does not repeat the top G or the bottom C, or you would hear the same note twice at each turn. Six and four make ten. Check it on the arpeggiator: it shows how many notes there are in one pass.

An arpeggiator breaks up the chord you are holding. Check that you can take the chords of the track straight away and whole: the course names a chord, and you play it in any inversion.

In the project the arpeggio comes in as a fifth track, with the "pluck" sound and an octave above the pad. It is the same harmony, only in motion, and it needs a floor of its own. The pad holds the chord underneath, and the pluck runs through it on top. Try the patterns and sounds, and listen for the point where the arpeggio starts to get in the melody's way.

The "Sounds to the Notebook" button hands the sounds you chose today to the tracks of your Notebook: melody, bass and chords.

17:00. Dynamics and velocity

So far every note in the project plays at almost the same strength. That is how a machine plays, and you can hear it: an arpeggio at a single velocity ticks like a metronome. Musicians call loudness and its changes dynamics, and in MIDI every note has its own velocity, from 1 to 127. In the piano roll of any DAW there is a velocity lane under the notes, like this one.

Press Play and switch between the presets. "Machine" puts every note at 100: even and lifeless. "Accents" makes the beats stronger and the rest softer, and the even stream of sixteenths at once falls into groups of four you can count. "Crescendo" goes from soft to loud over a bar, which is how electronic music builds up to a chorus. "Human feel" is the accents with a little random spread, the way a person plays.

Watch the line under the slider. On this synth velocity changes not only the loudness but also the filter cutoff: with the "Velocity → brightness" control at 2 octaves, a loud note is noticeably brighter than a soft one as well. Turn the control to zero and the accents stay, but they go flat, as if someone were just riding the volume. That is the difference between dynamics and a track's level: the volume fader on a mixer changes everything at once, while velocity changes each note and its sound.

Now play it yourself: the course names a dynamic, and you press any key hard enough, or softly enough, to hit it.

18:00. Song form

The four-bar loop is ready. But repeat it for three minutes straight and it wears thin by the third round, however good it is. A track is made of blocks, and each has its own job. The order of those blocks is the song form.

• The intro brings the listener in: only a few layers, often without bass or melody.
• The verse tells the story: it is simpler and quieter than the chorus, with a lower melody.
• The chorus is what it was all for: every layer plays, the melody is higher and louder, the bass and the beat are busier.
• The bridge comes once, towards the end: different chords, a different line-up. It is there for contrast, so that the last chorus sounds like coming home.
• The outro closes the track: the layers drop out and the sound fades away on the tonic.

Below is the form builder. At the top is a view like the one in a DAW: a row for each role, a column for each block, and whatever plays is filled in. Add blocks with the buttons, reorder them by dragging the ⠿ handle (or with the arrows), and switch layers on and off in each block, with the buttons on its card or straight on the cells at the top. Play plays the whole track.

The "Pop" template has eight blocks: intro, verse, chorus, verse, chorus, bridge, chorus, outro. At 100 BPM that is 28 bars, a little over a minute. In a real song the verses and choruses are usually twice as long, but the structure is the same. Look at the view at the top: the blocks differ not in their notes, which are almost the same everywhere, but in which layers are filled in. The verse has no arpeggio, and the bass plays whole notes. In the chorus everything comes in, and the bass switches to the kick's rhythm. In the bridge the drums and the melody drop out, and the chords change: F, G, E minor, A minor.

The bridge is older than rock. In American popular songs from the mid-1920s into the 1950s the dominant form was AABA: four sections of eight bars, where the third, B, is the bridge. That is why it is also called the middle eight. From the 1960s on, songs were more and more often written in verse–chorus form, and the bridge kept its place there: once, towards the end, for contrast.

The "Lay the Notebook out in this form" button takes your own song from the Notebook and repeats it once for every block in the form, switching its tracks on and off to match each block's layers. If the Notebook is empty, the form is laid out on "Ode to Joy".

Now by ear. A short track of several two-bar blocks plays. Tap the block names in the order you heard them. The ▶ next to each name lets you hear that block on its own.

19:00. Repetition, contrast and layers

The form builder has a template called "Layer by layer". It has no verse and no chorus: the same block seven times in a row, and each time one track comes in or drops out. The pad, then the arpeggio, then the bass, then the beat, then the melody, and back again. Listen: the track stays alive, though the notes never change once. The ear follows what has arrived and what has left.

These are the two main tools of arrangement. Repetition makes a track recognisable: the listener remembers what comes back. Contrast keeps them from getting bored: something has to change. You do not need a new melody for that. Adding a layer, taking one away or changing one sound is enough. The builder warns you if three identical blocks stand in a row with no change of layers.

Electronic music took from both. The climax of a dance track is a wall: every layer in, the filter wide open. The build-up to it is minimalism: layers one at a time, the filter opening little by little, eight bars of the same thing with one new sound.

The last exercise of the day is about layers. The whole track plays for two bars, then two more without one of its tracks. Which one?

The whole track plays for two bars, rhythm, bass, harmony and melody, then two more without one of its tracks. Which one is missing?

Evening: what you have made

In one day the mud has grown into a track. You split it into roles and gave each its own floor and its own sound: the beat at the edges of the spectrum, the bass at the bottom in the kick's rhythm, the pad in the middle, the pluck and the lead on top. You saw one part cover another, and a low-pass filter make room without a single new note. You switched on the arpeggiator, drew the dynamics, built a form from blocks and heard that contrast comes from layers.

The track is ready, but it lives in this browser. A real studio is a DAW, and the track has to be taken there whole: notes, tracks, tempo, form. How to do that, what survives the move and what stays behind, and how to check a track before you release it is the subject of the last chapter, chapter 21.

Each track in the Notebook now has its own sound: the "Sounds to the Notebook" button in the arpeggiator section hands it the sounds you chose today. The "Lay the Notebook out in this form" button turns your song into a track with blocks, and the rows of cells under each track show which bars it plays in. In chapter 21 the Notebook opens out into a full studio, and the track goes out as a MIDI file to any DAW.