Layering works when each layer owns something the others do not. The three axes that matter are frequency range, time, and stereo position — separate your layers on at least one of them and they add up. Overlap on all three and they cancel, mask, and sound smaller than a single sample.
Two sounds occupying the same frequency band do not sum to double the volume. They sum according to their phase relationship, which for real-world material is somewhere between full addition and partial cancellation. Two kicks both centred at 55 Hz will often be quieter together than either alone.
Masking is the second problem. When a loud sound and a quiet sound share a band, you stop hearing the quiet one entirely — you just hear a thicker version of the loud one, and you have spent CPU and headroom for nothing.
Give each layer a band and remove the rest. For a layered kick:
- Sub layer: low-pass at around 100 Hz. This is the weight you feel.
- Body layer: band-passed roughly 100–400 Hz. This is the thump.
- Click layer: high-pass at around 2 kHz. This is the definition on small speakers.
The same logic applies to bass (sub below 100 Hz, growl 200–800 Hz, presence 1–3 kHz) and to pads (one layer for the 200–600 Hz warmth, one for 4 kHz upward shimmer).
Cut steeply. A 24 dB/octave filter creates a genuine handoff; a 6 dB/octave slope leaves both layers in the crossover region still competing.
Attack and sustain are perceptually separate events. Take the first 20–40 ms from one source and the tail from another.
A snare built this way — a tight acoustic crack for the transient, a longer noisy layer for the body, a room sample for the tail — reads as one instrument because the ear groups sounds that start together.
The rule: only one layer should have a fast attack. If three layers all have sharp transients, you get flam and phasiness rather than punch.
For synth layers, detune by 5 to 15 cents — enough to create movement, not enough to sound out of tune. Beyond about 25 cents it reads as a chorus effect or an error.
Offsetting one layer by 5–15 ms also widens a sound, but only if you check mono. Larger offsets start sounding like a flam or a slap delay.
Keep the low layer mono and spread the upper layers. Sub content below roughly 120 Hz should stay centred — stereo bass loses energy in mono playback and eats headroom on a master. Put the width where the ear can actually localise it, above 300 Hz.
| Check | How | What you want |
|---|
| Phase | Flip polarity on one layer | The correct position is noticeably fuller |
| Mono | Sum to mono, listen | No layer disappears or thins out |
| Solo each layer | Bypass the others | Every layer has an identifiable job |
| Level match | A/B against a single sample | Layered version is not just louder |
Start with the layer that carries the most emotional weight — usually the one with the most character — and build around it. Add the second layer, then immediately cut the frequency range it does not need. If you cannot articulate what the third layer adds, delete it.
Bounce the layered result to a single audio file once you are happy. Working with a committed sound stops you from endlessly re-tweaking a five-layer stack and forces arrangement decisions instead.
If you are using stems separated from an existing track as layer material, remember that owning your export never gives you rights in someone else's recording — that is a clearance question, not a technical one.
Related reading: mixing bass, stereo width and panning, and how to fix a muddy mix.
Start making music in Veena
Free, browser-based, no downloads required.
Try Veena Free