Sound Design Basics: Building Sounds From Scratch
Most producers build tracks from sounds someone else recorded. There is nothing wrong with that, but it has a ceiling: you can only use what exists, and you can never be sure who else is using the same thing. Synthesis removes both limits. This article explains how a sound is built from nothing — the physics behind it, the three techniques that cover most of what you will need, and why a synthesised sound never needs clearing.
What a sound actually is
Every sound is three things happening at once: which frequencies are present, how loud each of them is, and how all of that changes over time.
That last part does most of the work. Take a piano note and cut off the first thirty milliseconds — the hammer hitting the string — and most people can no longer tell you it is a piano. The frequencies did not change. The attack did. This is why sound design is mostly envelope work, not filter work.
So when you build a sound from scratch, you are answering three questions. What is vibrating? How is it being hit or excited? And how quickly does that vibration die away?
Technique 1: modal synthesis — for anything solid
Hit a metal plate and it rings at a set of specific frequencies at once. Those are its modes, and their ratios are fixed by the physical shape of the object. A string has modes at 1×, 2×, 3× the fundamental — whole numbers, which is why strings sound musical. A cymbal has modes at ratios that are nothing like whole numbers, which is why it sounds like noise with a pitch hiding inside it.
Modal synthesis means stacking sine waves at those ratios and giving each one its own decay time. Higher modes fade faster, because in a real object high frequencies lose energy quicker. Get the ratios and the decay curve right and it sounds like a real object being struck — without a microphone ever being involved.
This one technique covers bells, metal, wood, glass, tuned percussion and the body of almost any drum.
Technique 2: Karplus-Strong — for anything plucked
This one is almost absurdly simple, and it sounds far better than it has any right to. Fill a short delay line with noise. Feed the output back into the input, and average each sample with the one before it as it goes round.
The averaging is a low-pass filter. Each pass round the loop, the high frequencies lose a little more energy than the low ones — exactly what happens on a real string. Within a few dozen milliseconds the noise burst has turned into a pitched, decaying pluck. The length of the delay line sets the pitch. How much averaging you apply sets how bright and how long the decay is.
Guitars, harps, kotos, rubber bands, anything that gets plucked or struck and rings. It costs almost nothing to compute, which is why it turns up in hardware from the 1980s onwards.
Technique 3: filtered noise — for anything with air or friction
Wind, breath, cymbals, rain, snare rattle, distant traffic. These are not pitched, so the previous two techniques do not apply. What they have in common is noise shaped by a resonance.
Start with white noise, which contains every frequency equally. Then carve it: a band-pass filter around 4 kHz gives you a hiss with a character; sweeping that filter down over half a second gives you something passing by. The envelope decides whether it reads as a hit or as an atmosphere.
Most real sounds are combinations. A snare drum is a modal body — a tuned shell — plus filtered noise from the wires underneath. Get the balance between the two right and you have a snare that no sample library contains.
The transient matters more than the tail
If you take one practical thing from this article, take this: spend your time on the first fifty milliseconds.
Listeners identify a sound almost entirely from its attack. The sustain and release carry the mood, but the transient carries the identity. A kick drum that reads as weak usually does not need more low end — it needs a sharper click at the top of the attack so the ear can locate it.
When a synthesised sound feels fake, the attack is usually the reason. Real objects do not start instantly and they do not start cleanly. A tiny burst of noise in the first few milliseconds — the sound of the hammer, the stick, the fingernail — is often the difference between synthetic and convincing.
The part nobody mentions: clearing
If you sample a record and release the track commercially, you need permission from both the owner of the recording and the owner of the composition. That is two clearances, each of which can be refused or priced out of reach after your track is finished.
Royalty-free sample packs solve most of this, but not all of it. A pack is only as clean as its source, and packs built from recorded material occasionally contain things the maker did not have the right to sell on.
A synthesised sound has no such history. It was calculated, not recorded. There is nothing to clear because there is no underlying recording and no underlying composition. For anyone releasing commercially, that is not a technical detail — it is the difference between a track you own and a track that could be pulled.
Where to start
You do not need new equipment. Every major DAW ships with a synth capable of all three techniques.
A reasonable first exercise: build a hi-hat. Take white noise, band-pass it somewhere around 8 kHz, and put a very fast decay envelope on it — under 60 milliseconds. That is a closed hat. Lengthen the decay to 300 milliseconds and you have an open one. Ten minutes of work, and you now have hats that exist nowhere else.
Then do a kick: a sine wave whose pitch drops sharply from around 120 Hz to 45 Hz in the first 40 milliseconds, with a short click layered on the attack. That pitch drop is the entire trick, and it is the same trick behind almost every electronic kick you have ever heard.
Common questions
Do I need an expensive synth to do this?
No. The stock synths in FL Studio, Ableton and Logic all handle every technique described here. What limits results is understanding what you are building, not what you are building it with.
Is a synthesised sound as good as a recorded one?
Different, rather than better or worse. Recordings carry the accidents of a real room and a real performer, which are hard to fake. Synthesis gives you total control and no legal history. Most professional productions use both.
Can I sell music made with synthesised sounds?
Yes, without restriction. There is no source recording and no source composition, so there is nothing to clear and no one to pay.
How long before this starts sounding good?
Usable drums within a few hours. A recognisable sound of your own takes considerably longer, because that is a matter of taste rather than technique.
Rather start from finished sounds?
Our sample packs are built the way this article describes. Royalty-free, WAV, ready for any DAW.