making a song
drafting a section
making the arrangement
- sing out the idea in one audio file (sing your song arrangement)
- this should include all the prominent elements that you want to have
- turn that general idea into midi
- sing into pitch finder
- from that midi, get the key scale (making a key scale)
- turn some of your midi into chords (making chords)
- general tempo? very slow, slow, normal, fast, very fast
- rubato?
more
- adding another instrument
- typical ones to choose from
- drums
- bass
- harmony
- sing it on top
- typical ones to choose from
- make the groove for those
- each of those should have a specific pattern that fits the groove of the song
- find a reference track. every 30 minutes listen to it while building the song
- use a ton of automation
- this gives movement
structure a song before you compose details
drafting general song
pieces you can do:
- foreground elements
- different types of fg
- lead melody
- repeating pattern melody
- ex: “no surprises; radiohead”
- ex: an arpeggiator
- n: when you do an arp, do not do the algorithmic option that does it for you. do it in midi this way you can adjust humanization, velocity, note length, etc
- “drone” melody (do not count toward the 3 foreground, they count more as harmony)
- melodies that just follow a note in the chords (ex: some basslines just do long notes for the root note)
- however, if it’s following a note in a chord, but has groove (ex: it’s not just doing a drone note for that chord), then it is a PATTERN (if it follows the same groove over time), or a LEAD (if it’s groove changes over time)
- fx
- evil laugh
- impact/hit
- single dobro slide guitar sound over lyrics every end of bar.
- vocal fill “whoop!”
- a riser
- can have <= 3 foreground at once
- fx tracks do not “stack”. ex: if you have 10 fx tracks at the same time, it only counts as 1 foreground element. because it is registered as 1 sound in the listener’s brain
- ex: if you have a loud gunshot spree sfx, for the portion where that plays, you can only have 2 other foreground
- if it’s the same type (lead or pattern), they must be in different pitch ranges
- low, mid, high—these are the ranges
- if they are different types, they are allowed to be in the same pitch range
- ideally, each is panned differently and has a different sound
- fx tracks do not “stack”. ex: if you have 10 fx tracks at the same time, it only counts as 1 foreground element. because it is registered as 1 sound in the listener’s brain
- different types of fg
- percussion (unpitched, if it’s pitched, i consider it melodic)
- drums
- kick
- snare
- hi hat
- normal (closed)
- open
- cymbal
- shaker
- claps
- drums
- harmony (stuff playing the actual chords or something very similar)
- ambience—to ground the song in a space and fill emptiness
- ex: tape hiss
- ex: street noises
- ex: drones
overall idea
purposefully break >= 1 rule, follow the rest
as short as possible, but no shorter
- generally, this means less than 3:00
doesn’t sound basic
build up throughout song, deliver with climax
the song should have at least 3 distinct “sounds”
- at at least 3 different parts of the song, the overall sound should sound different
end quickly
dont have long portions with no lyrics (>10 seconds)
super emotional
if you have a chorus, the lyrics and sound should be slightly different every time
- ex: 1. “gravity … it’s working against me”. 2. “gravity … stay the hell away from me”
something new must happen every <= 7 seconds
- ex: a new element is introduced
- ex: the song is pitch shifted for a second
- ex: a start-stop silence
dont introduce two new elements at the same time, unless you are doing a big explosive moment (ex: me and your mama childish gambino)
generally, the hook should arrive within the first 30 seconds
making a track
(a track is one track in the song)
what instrument?
what frequencies do i want for this sound?
- are those frequencies already being taken up by other sounds?
- eq out everything else
- high/low pass and/or shelves
how long/constant should the sound be?
- ideally, they should be rather short. you kind of want stuff to ocme in waves, rather than just constant
- so basically cut out parts of the sound that are not necessary. let it default to empty space
- the more you shorten a sound, the more you create space for the next sound to shine and sound good
- ex: a half-second kick is better than a full-second kick, because it leaves space for other stuff, and you only really need the transient
- should be like 1/8 or 1/16 note
- ex: waves of bass, rather than constant