Here’s what most producers do with Serum 2 in my experience: cycle through the five oscillator engines, hit Spectral, hold one note, shrug, move on. Fair enough. The Serum 2 spectral oscillator has an intimidating face, and “spectral resynthesis” sounds like something from a DSP lecture hall.
The concept underneath is actually simple though. Get it once, and suddenly almost any sample – a spoken word, a piano note, a coffee mug you tapped with a spoon – becomes a playable instrument you can morph and resculpt in ways no sampler will ever manage.
Monosounds has been shipping Serum preset packs since 2022, and since the 2.0 update this engine has carried more of my patches than anything besides wavetable. So here’s what it really does, explained in normal words, with three recipes you can copy today.
Spectral Resynthesis in Plain Words
Any pitched sound is sine waves stacked on top of each other. One piano note isn’t a single wave. It’s a fundamental with dozens of overtones riding above it, each at its own level, all shifting as time passes.
Spectral resynthesis takes a sample apart into those sine components – partials, in Serum 2’s vocabulary. It then reconstructs the sound live from the partials rather than playing the recording back.
That’s genuinely the whole idea. A sampler gives you a frozen photograph of a sound. Resynthesis gives you the recipe instead. And recipes can be messed with: pull ingredients out, turn others up, stretch the timing, reorder the steps.
Everything here happens in the frequency domain. You’ve stopped editing audio; you’re editing harmonics now. That’s exactly why this engine pulls off things a regular sampler flatly cannot.
How the Serum 2 Spectral Oscillator Works
Flip any of the three oscillators over to the Spectral engine, then drag a sample in. Serum 2 breaks the audio down into spectral frames – moment-by-moment snapshots of the partial content across the file.
Three abilities matter more than everything else from there.
Muting and Isolating Partials
The sound becomes editable partial by partial. Kill the bottom five and the fundamental is gone, leaving nothing but the shimmer above. Kill every even harmonic instead and any source turns into a hollow, clarinet-ish tone.
The move I love most: solo a thin band of upper partials out of a piano. What remains has nothing piano about it. It sounds like glass.
Morphing Between Frames
Frame position behaves like WT Pos on a wavetable, only here the frames trace the timeline of your sample. Sweep through slowly and a two-second vocal stretches into a two-minute evolving pad. Park on a single frame and that exact instant drones forever.
Put an envelope or LFO on frame position and the sound starts moving by itself. About 80 percent of the magic sits right here.
The Transient Problem
The honest catch: attacks get smoothed away by resynthesis. Sine waves handle sustained tone brilliantly and clicks badly. A resynthesized piano sheds some of its hammer, and a pluck put through this can go mushy on you.
Two ways out. Fake a sharper attack with a fast amp envelope, or put the raw sample on a second oscillator in the Sample engine, covering only the first 50 milliseconds. Nearly every spectral patch I ship uses that layering fix.
Where Spectral Fits Among the Five Engines
Each oscillator in Serum 2 offers five engines, and each engine has its lane. The full breakdown lives in my guide to all five Serum 2 oscillator types, but here’s the Serum 2 spectral oscillator’s spot in the lineup:
| Engine | Best for | Weak spot |
|---|---|---|
| Wavetable | Classic basses, leads, EDM staples | Static without heavy modulation |
| Sample | One-shots, exact playback, attack layers | It’s a fixed recording |
| Multisample (SFZ) | Realistic keys, strings, mapped kits | Bigger memory footprint |
| Granular | Gritty textures, stutters, motion | CPU climbs fast with grain count |
| Spectral | Morphing pads, harmonic surgery | Smears transients |
Granular is the nearest cousin, and the two get confused all the time. Granular slices the audio into tiny time-fragments and replays them; spectral reconstructs everything from sine partials. Grit and texture stay with granular, pitch and smoothness stay with spectral. For the grain side, see my Serum 2 granular tutorial.
Three Sound Ideas You Can Build Today
All three recipes below run on the Serum 2 spectral oscillator, stock FX, and a single sample you record yourself or already have lying around. Nothing exotic required.
1. Airy Keys from a Piano Sample
Begin with one held piano note – C3, three or four seconds long, recorded without reverb.
- Osc A: Spectral engine, pull the piano note in.
- Silence the bottom five partials. Out goes the fundamental, the overtone shimmer stays behind.
- Route Env 2 to frame position with attack near 800 ms, so the harmonics open up after the note has started.
- Unison 3, detune 0.08.
- Osc B: Sample engine, the identical piano note, sitting around -12 dB, purely to restore a hint of hammer.
- FX: hall reverb, size near 60, mix near 35.
What comes out sounds like a piano being remembered in a dream. Several key and pad patches in our free Serum 2 synth presets run variations of this, so a finished version is there to hear.
2. Ghost Vocal Pad
Sing or speak one word, about a bar long at 120 BPM, and record it. A phone recording is fine – resynthesis swallows a surprising amount of noise.
- Osc A: Spectral engine, load the vocal.
- Route LFO 1, triangle, synced to 8 bars, into frame position at roughly 35 percent depth. The word melts apart and slowly reassembles.
- Silence the noisy partials at the very top to tame sibilance – for a pad, that highest range contributes basically nothing.
- Unison 5, detune 0.12, width full.
- Amp envelope: attack 1.2 s, release 2.5 s.
- FX: one of Serum 2’s new reverbs, big and dark, with a gentle chorus after it.
Half the atmosphere patches in our free pack of 250+ Serum 2 presets use this exact trick. Load one, click into the oscillator, and the muted partials are right there to inspect.
3. Metallic Pluck
Whack something made of metal – a pan, a radiator, a mug – and capture a second of it. Metal rings with inharmonic partials, which spectral analysis catches perfectly and a plain wavetable simply can’t hold.
- Osc A: Spectral engine, drop the hit in.
- Isolate a partial band, somewhere around the 12th through the 30th. The clang lives in that cluster.
- Env 1 on amp: attack 0, decay 400 ms, sustain 0.
- Also send Env 1 to frame position, just a little, giving every note a small spectral sweep.
- FX: Bode frequency shifter at +7 Hz for added metallic detune, followed by a 1/8-dotted ping-pong delay.
Plucks like this drop beautifully into future bass and melodic house. For ready-built examples, the pluck and chord patches in Harmonic Horizons, our future bass Serum preset pack, are built on these same ideas.
Keeping the CPU Bill Reasonable
Spectral costs more than wavetable, no way around it. Rebuilding sound out of dozens of sine partials burns cycles, and unison stacked on top multiplies the bill.
My working setup: unison at 3-5 rather than 16, polyphony capped at 8 for pads, oversampling parked at 2x unless aliasing is actually audible. Then freeze or bounce the track the moment a part is written. Xfer keeps official CPU guidelines on the Serum 2 page if you want their full checklist.
One extra trick worth knowing: when a spectral patch is done and the live morphing no longer matters, run Resample to Oscillator and bake it into a wavetable. The tone stays, the CPU cost falls away, and you can keep carving – that workflow is covered in my post on making custom wavetables in Serum 2.
Serum 2 Spectral Oscillator FAQ
Is the spectral oscillator the same as granular?
No. Granular fires tiny time-slices of the audio, up to 256 grains simultaneously, and preserves the raw texture. The Serum 2 spectral oscillator rebuilds everything from sine partials instead, giving smoother output plus per-harmonic editing. Feed both the same sample and you get very different results.
Can I use any sample in the spectral oscillator?
More or less. Any WAV dragged onto the oscillator gets analyzed. The best resynthesis comes from clean, pitched, mostly-mono material. Dense mixes and full drum loops just smear into mush – point that material at the Sample or Granular engine instead.
Why does my spectral patch sound washed out?
Two likely reasons. Resynthesis smooths transients away, so put the raw sample underneath as an attack layer. And a pile of quiet upper partials creates a hiss-like blur – silence the top range and focus usually snaps back immediately.
Does Serum 1 have a spectral oscillator?
No. It showed up with Serum 2 in March 2025. Since the upgrade is free for Serum 1 owners and old presets carry over, sticking with version 1 makes no real sense.
Steal Our Patches First
Reading isn’t the quickest path into this engine. Reverse-engineering is. For a shortcut, take our free 250+ Serum 2 presets – genuinely no strings. Open the spectral patches, note which partials got muted and what drives frame position, and dismantle them until it all makes sense.
I run Monosounds.studio and make the packs myself. Ten-plus years of producing, 97+ releases, 16,000+ downloads. Every Serum 2 preset we ship passes a CPU check on a mid-range laptop and gets used in a real track before release. If it does not survive a mix, it does not go in the pack.
More about Maxim & the studio →Looking for the packs themselves? Here is the full Serum 2 preset catalogue — sorted by genre and by sound, with a demo on every one.
