Ask any producer and they’ll show you a folder of vocal chops. And most of them are still bouncing every chop out of a sampler one at a time before a synth ever sees it. Five years ago that workflow made sense. Today it doesn’t.
Serum 2 builds a Sample engine, a Multisample engine, granular, and spectral resynthesis into every oscillator. Which means serum 2 vocal chops can now live inside the synth itself: pitched, filtered, modulated, and resampled without ever leaving the plugin.
At Monosounds we’ve been making vocal patches in Serum 2 since it landed in March 2025. These are the five methods I genuinely use, with the settings I begin from every single time.
Serum 2 Vocal Chops Start in the Sample Oscillator
This route is the quickest. Grab a short, dry vocal phrase – a bar or two, no baked-in reverb. Flip OSC A’s engine from Wavetable to Sample, then pull your .wav onto the oscillator. The order is important: drop the file while the oscillator still sits in Wavetable mode and Serum turns it into a wavetable instead. That’s sometimes exactly what you want. Just not yet.
Two settings need attention immediately. Keytracking first. Switched off, the vocal plays at its recorded pitch on every key, which is the behavior you want for one-finger chop stabs. Left on, the sample repitches chromatically, the classic playable-chop sound. Then there’s the sample start point. Nudge it forward to where a vowel begins and you’ve got an entirely new chop without cutting a thing.
Time to add movement. A fast-attack envelope on the amp with decay near 200-400 ms turns every key press into a tight stab. Then route an LFO or a macro into sample start position, so each note pulls a different slice of the phrase. That single modulation route covers about half of what a sampler used to do for me.
Map Chops Across Keys with Multisample
Stretch one sample across a whole keyboard and it collapses quickly. Beyond four or five semitones the vowels smear, the consonants drag out, and you’re firmly in chipmunk territory. Most serum 2 vocal chops that come off cheap come off cheap for precisely this reason.
The Multisample engine solves it. It reads SFZ files, so a different chop can sit in every key zone. How I work: slice 8-12 chops out of one vocal take, keep the files in a single folder, and assign each one its own key. So C3 becomes an “oh”, D3 a “hey”, E3 a reversed breath. The chops play like drums, with every zone staying inside its natural pitch range.
SFZ looks scary and isn’t – it’s plain text, a minimal mapping runs a few lines per sample, and free tools exist that build one from a folder. The playable patches in our GO2 beatmaking pack run on the same logic: a mapped patch outlives every new project, while a folder of bounced chops is forgotten within three months.
Formant Shifting Without the Chipmunk Effect
Repitch a vocal and its formants travel with the pitch. Upward sounds tiny, downward sounds like a monster. Sometimes that IS the point – half of future bass is built on chops pitched up three or four semitones, and if that’s where you live, my future bass in Serum 2 tutorial handles the chord side of the sound.
When you need pitch movement minus the cartoon voice, though, Serum 2 offers two solid paths. Path one: go ahead and make the vocal a wavetable after all. Drop the .wav onto an oscillator in Wavetable mode, then reach for the Formant warp mode. Warp moves the vowel character while the played note handles pitch on its own. The result leans synthetic rather than natural, but that robotic vowel tone became a genre staple for good reason.
Path two is the Granular engine, which severs the pitch-time link entirely. It’s earned its own section.
Granular Vocal Textures
Switch the oscillator engine to Granular and pull in the same vocal. Rather than playing the file front to back, Serum 2 sprays tiny overlapping grains of it – as many as 256 at once. Pitch and playback position go independent, so a chop can sit at any pitch, held as long as you like, with no stretch artifacts.
Where I begin for a vocal texture: grain size near 80-150 ms, density high enough that grains overlap, and a slow LFO wandering the grain position across the vowel part of the phrase. Sprinkle a little random spread onto the position and a static one-shot becomes a breathing pad. Tune a second granular oscillator up an octave at low volume and it comes across as ghostly harmony.
Stacked vocal layers of this kind are exactly the sort of source the granular engine likes.
Mind the CPU here. Dense grains combined with big unison gets expensive in a hurry. My rules: unison at 1-3 on granular oscillators, oversampling down to 2x while sketching, freeze the track once the part is locked in. For the engine explained control by control, my full Serum 2 granular tutorial goes far deeper than one section can.
Spectral Ghost Vocals
The Spectral engine resynthesizes a vocal into partials: individual sine components you can mute, edit one at a time, or morph between frames. The ghost-vocal sound lives here. Resynthesize a sung phrase, thin the partials out, drag the frame movement way down in speed, and what emerges is unmistakably a voice and unmistakably not a human one.
My standard move: cut the lowest partials, hold onto the region from roughly 300 Hz to 3 kHz where the vowel information sits, and morph across two frames over four bars. Under sparse drums it works absurdly well. The whole engine gets a proper teardown in the Serum 2 spectral oscillator guide if you want the complete picture.
Which Engine for Which Job
My decision table, patch by patch:
| Engine | Best for | Pitch behavior | CPU cost |
|---|---|---|---|
| Sample | Clean one-shot chops, stabs | Repitches whole sample | Low |
| Multisample (SFZ) | Many chops mapped to keys | One chop per key zone | Low-medium |
| Wavetable + Formant warp | Robotic vowel leads | Pitch and formant separate | Low |
| Granular | Pads, stretched textures | Pitch and time decoupled | Medium-high |
| Spectral | Ghost vocals, morphing vowels | Partial-level control | Medium-high |
No Vocal Sample? Fake One with Formant Filters
A convincing vocal-ish synth needs no recording at all. Vowels are nothing but resonant peaks called formants, and recreate those peaks and the ear fills in a voice.
The build: one saw oscillator with gentle unison – unison 3, detune around 0.08-0.12 – feeding two band-pass filters in parallel. Serum 2’s dual filter routing handles this natively, no outside EQ required. Set the first filter somewhere in the 300 to 700 Hz range and the second between 900 and 2300 Hz, with resonance high enough that each nearly whistles. Then put both cutoffs on one macro moving between vowel positions, so a single knob glides from “oo” through “ah” to “ee”.
Two final touches close the deal. A vibrato LFO on pitch at 5-6 Hz, tiny depth, delayed onset, because no real singer starts vibrato from the first instant. And a Bode frequency shifter on its own FX bus, offset by only a few Hz, supplying that faintly detuned unnatural-voice character. Rather reverse-engineer than build? Our 250+ free Serum 2 presets contain vocal-style keys, atmospheres, and chord patches ready to be opened and studied.
Serum 2 Vocal Chop Questions
How do I install vocal chop presets in Serum 2?
A pack shipped as a .SerumPack file goes in through Serum 2’s Menu tab via Import Preset Pack. Anything else, copy into your Serum 2 presets folder – the browser picks up subfolders, so keep packs tidy. Serum 1 presets load without issues too. Our Serum 2 preset guide lays out which packs deserve your attention first.
Can Serum 2 time-stretch a vocal to my project BPM?
Kind of. The Sample engine ties speed to pitch, old-hardware-sampler style. The Granular engine unties them, and for chops under a bar or two it does the job beautifully. To stretch a full vocal onto a new tempo, I still handle it in the DAW first and load the result – cleaner sound, lighter CPU.
Why do my serum 2 vocal chops sound thin?
The source is usually to blame, not the synth. A chop captured with room reverb or heavy processing fights every one of these engines. Begin dry, high-pass near 100-150 Hz, and park Serum 2’s reverb on a separate FX bus so the tail doesn’t retrigger with each note. A quiet saw layered an octave under the chop also fattens things up fast.
Is Serum 2 a paid upgrade for these vocal features?
No. Serum 2 comes free to Serum 1 owners with lifetime free updates, currently sitting at version 2.1.x. All five oscillator engines ship included, with nothing behind a paywall. Xfer’s product page has the details.
Start with a Phrase You Already Have
Take whatever vocal is sitting in your project folder right now, rough takes included. Sample engine to begin, granular once it needs motion, spectral once it needs to get weird. Between them, those three handle 90% of serum 2 vocal chops in real tracks; taste covers the rest.
And for a shortcut, pick up our free 250+ Serum 2 presets – royalty-free, nothing attached. Tear the vocal-ish patches open and steal the wiring.
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 →
