AU · VST3 · StandaloneMac & Windows8 modules · one window10.7 ms tracking mode
What it is
Eight modules. One window. Built for the voice.
CLEANSLATE is a complete vocal chain in the order a professional session
runs it: noise suppression, cleanup, spectral resonance control, tape
saturation, de-essing, a surgical EQ and two stages of compression, in a
single window with the signal flowing top to bottom.
Every module is modelled on the analogue compressors and spectral
processors engineers reach for, then tuned specifically for vocals rather
than shipped as a general-purpose effect. Spectral knows the regions where
a voice builds resonance and listens for them in your take; if it finds
nothing, it removes nothing. The compressors, the noise suppression and
the de-esser follow the same principle: prepared for the voice, acting
only when the voice calls for it, with every parameter still under your
control.
Switch on TRACK and the chain drops to 10.7 ms of latency, so the
same processing that finishes a vocal can be sung through live.
FormatsAU · VST3 · Standalone
SystemsmacOS 11+ (Apple silicon & Intel) · Windows 10+
Tested onA wide range of microphones, dynamic to condenser
LiveTRACK mode · 10.7 ms latency
Signal flow · top to bottom
What's in the chain.
01
Noise
Adaptive suppression that kills the room, not the take. No learn button, nothing to arm — it just listens.
Tap for the full story
02
Cleanup
High-pass, low-pass and the mud filter. The stuff you do on every vocal, done before you have to think about it.
Tap for the full story
03
Spectral
Resonance control with four nodes and a live reduction burn. Boxiness, harshness and mic ring, carved out as they appear.
Tap for the full story
04
Tape
Oversampled tape saturation with Character, Steady, Focus and Tube. Level-invariant, so quiet takes get the same glow.
Tap for the full story
05
De-ess
Adaptive threshold, no attack, no release, no lisp. Wide or split, it only ever touches the ess.
Tap for the full story
06
EQ
Twelve surgical bands, ten shapes, slopes to 96 dB, per-band dynamics. Air Link lifts the top as the de-esser works.
Tap for the full story
07
Catch
A FET compressor in the classic fast style: grabs the peaks, adds the grit, holds the vocal in front.
Tap for the full story
08
Level
Opto levelling after the catch. Slow, musical, invisible — the reason the vocal never leaves the mix.
Tap for the full story
CLEANSLATE
Made by engineers and vocalists, for engineers and vocalists.
The boring bits
Runs everywhere.
Formats
AU, VST3, Standalone app
DAWs
Logic, Ableton Live, FL Studio, Cubase, Studio One, Reaper, Bitwig, Luna, GarageBand — anything that loads AU or VST3
Systems
macOS 11 or later, universal binary. Windows 10 or later, 64-bit
Latency
~95 ms full chain, reported to the host. TRACK mode: 10.7 ms for live monitoring
Loudness match
1:1 button — the output lands at the input level whatever the chain did
Licence
One key, two machines, activate once then work offline. Move seats yourself
Delivery
Key and installers by email the moment payment clears
Updates
One update a month, free with your key
Made by
BADSCANDAL Audio — engineers and vocalists
CLEANSLATE · v1
Get the chain.
One licence. Every DAW you own. The whole vocal chain, in the order it
should run, in one window. Pay once and the key is in your inbox before you've closed the tab.
Instant key by email2 machinesWorks offlineMonthly updates
Installers are free to download. The key unlocks them.macOS·Windows
Questions
Things you'd ask.
Will it load in my DAW?
If your DAW loads AU or VST3, yes. Logic, Ableton, FL, Cubase, Studio One, Reaper, Bitwig, Luna and GarageBand are all covered. Pro Tools (AAX) isn't at launch — if you need it, tell us and it moves up the list.
How does the licence work?
You get a key by email. Enter it once on each machine (up to two), and from then on the plugin works with no internet at all. Deactivate a seat yourself when you change computers. Until a key is entered the plugin passes audio through clean, so a session never goes silent.
Is there a trial?
Not yet. Everything on this page is the real plugin, recorded through it. If it doesn't do what it says on a vocal you own, write to us within 14 days and we'll sort it.
Do I get updates?
Yes. We treat a purchase as something to maintain and improve, not something to sell and forget. CLEANSLATE receives one update every month, refining the existing modules and adding new features, with a single aim: the most efficient all-in-one vocal chain for producers and artists.
Updates are free, never ask for your key again, and the plugin tells you when one is ready: press CHECK FOR UPDATE in its footer, or visit badscandal.com/updatecleanslate to see what changed and download the latest build.
That roadmap is shaped by the people using it. Send feature requests or bug reports to contact@badscandal.com and we will address them.
Where do I download it?
Right here, before or after you buy. The installers are free; the plugin passes audio through clean until a key is entered.
macOS 11 or later, Apple silicon and Intel, signed and notarised by Apple. Windows 10 or later, 64-bit. Your key also arrives with both links in the email.
01 · Noise
Kills the room. Keeps the take.
An adaptive noise suppressor that tracks the room's noise floor and subtracts it before anything else in the chain hears it.
What it does
It listens to the spectrum continuously and builds a model of everything that is not the voice: air conditioning, computer fans, hum, hiss. That model is subtracted between and under the words, and what remains in the pauses is a steady, natural room tone rather than a gate opening and closing. There is no learn button to press and nothing to arm mid-take.
Why it's better
The gain rule is the same class of speech-enhancement mathematics used in broadcast and forensic tools, tuned for sung vocals rather than speech. Held notes lose almost nothing (measured erosion of 0.04 dB on sustained tones), sibilance and breath onsets keep their attack, hum and whine that sit still for a few seconds are absorbed into the noise model, and a comfort-noise bed keeps pauses sounding like a room instead of a vacuum. Everything that is not voice gets quieter; the voice stays exactly where it was.
The controls
AMOUNT
How much of the noise between and under the words is removed. 100 is full reduction down to the ambience floor; 30 is a gentle clean-up.
AMBIENCE
How much room tone is left behind. Lower is drier; the remaining floor is a steady bed, so pauses never sound gated.
LISTEN
Monitors only what is being removed. If you can hear words in here, back off AMOUNT.
02 · Cleanup
Rumble and mud, gone before the tone starts.
A high-pass filter and a low-mid cut, placed first so every stage after them works on a clean signal.
What it does
The high-pass removes handling noise, proximity boom and stand rumble below the voice, at 12, 24 or 48 dB per octave. The mud cut is a single bell in the 150 to 500 Hz region where boxiness lives. Sweeping the corner glides instead of stepping, and changing the slope crossfades between filters so nothing clicks.
Why it's better
These are the two moves every engineer makes on every vocal, done in one place and done precisely: double-precision filters, a corner that glides instead of stepping, and slope changes that never pop in a session. Getting this out of the way early means the compressors and the tape are never fed rumble they would otherwise react to.
The controls
HPF
High-pass corner. Removes rumble, handling noise and proximity boom below the voice.
SLOPE
Steepness of the high-pass: 12, 24 or 48 dB per octave. Steeper is cleaner but less natural right at the corner.
MUD FREQ
Centre of the low-mid cut, where boxiness lives, roughly 150 to 500 Hz.
MUD CUT
Depth of the low-mid cut.
MUD Q
Width of the low-mid cut. Higher is narrower.
03 · Spectral
Resonances found and carved, only when they appear.
A dynamic resonance suppressor with four nodes that watches your vocal for build-ups and pushes them down as they happen.
What it does
You place up to four nodes on the regions where a voice tends to ring: low-mid boxiness, upper-mid harshness, mic and room resonances. Spectral listens inside each region and, when a peak stands out from the harmonics around it, dips exactly that peak by exactly as much as it sticks out. When nothing resonates, nothing is removed. The display shows the reduction being carved in real time.
Why it's better
Calibrated against the industry reference for this kind of processing, so depth, width and timing behave the way experienced engineers expect. Precision narrows each notch toward surgical, Selectivity leaves ordinary harmonics alone and catches only true resonances, and reduction saturates rather than chasing every peak, which is why it stays musical at high depth. A Delta mode plays only what is being removed, so you always know what you are giving up.
The controls
DEPTH
How far resonances are pushed down. Reduction saturates instead of chasing every peak, so it stays musical.
PRECISION
Width and maximum depth of each notch. Low is broad and shallow, high is narrow and surgical.
SELECTIVITY
How prominent a peak must be before it counts. Higher leaves ordinary harmonics alone and only catches real resonances.
ATTACK / RELEASE
How quickly suppression engages and lets go, in milliseconds.
STYLE
SOFT eases resonances down in proportion to how much they stick out. HARD lets absolute level push harder.
DELTA
Monitors only what is being removed. If you hear words in here, lower DEPTH.
04 · Tape
Warmth that ignores your level.
An oversampled tape stage with Character, Steady, Focus and Tube: harmonics, gentle squash and density, without touching the words.
What it does
Drive pushes the vocal into a tape model running at four times the sample rate, so the harmonics are clean rather than aliased. Character moves from velvet, a soft tube-like curve that is almost all second harmonic, to crunch, a hard knee with odd harmonics. Speed sets the head bump and the high-frequency roll-off together: 7.5 is thick and dark, 30 is clean and wide. Auto gain keeps the loudness where it was, so turning up drive adds colour, not volume.
Why it's better
Steady is the difference. At 100 a quiet verse gets exactly the same harmonics as a loud chorus, measured at the same distortion figure across a 24 dB range of input, with the level left untouched. Focus band-limits where the warmth is made, so the sub and the air pass around the tape and you can push drive without fizz. A transient guard lets consonants through cleaner than vowels, and the dry path is delay-matched to the wet, so any mix setting is phase-safe.
The controls
DRIVE
How hard the signal hits the tape, 0 to +24 dB. Harmonics, high-frequency squash and low-end density all grow with it. Auto gain keeps the volume steady.
CHARACTER
Velvet on the left: a soft curve biased like a class-A tube, almost all second harmonic, warm and round. Crunch on the right: a hard knee, odd harmonics, hotter and edgier.
STEADY
How much the warmth ignores your level. At 0 quiet passages stay clean and loud ones saturate. At 100 a quiet verse gets exactly the same harmonics as a loud chorus, with the level left untouched.
FOCUS
Where the warmth is made. At 0 the whole band is driven; turning up narrows it toward the body of the voice, so the sub and the air pass around the tape untouched.
TUBE
A soft tube stage in front of the tape, mostly second harmonic. Two gentle stages in series give a layered density one curve cannot; it never changes the level.
SPEED
Tape speed. Moves the head bump, the high-frequency loss and where the top starts to saturate together. 7.5 is thick and dark, 30 is clean and wide.
MIX
Parallel blend of dry and tape. Phase-aligned, so any setting is safe.
05 · De-ess
Catches the ess, never the word.
An adaptive de-esser with no attack, no release and no threshold to babysit.
What it does
A sibilance detector looks 10 ms ahead of the audio and only opens on the spectral signature of an ess or an affricate, never on a loud dark vowel. The threshold adapts to the performance and freezes during silence and during sibilance itself, so a quiet verse and a loud chorus are de-essed to the same degree. Range sets the most an ess can be turned down; Split dips only the sibilant band and leaves the body alone.
Why it's better
Most de-essers are compressors with a filter in front, and they lisp when pushed. This one approaches the range asymptotically, so loud esses stay proportionally louder than quiet ones and nothing ever goes dead. The lookahead catches the first millisecond of every ess. Air Link feeds its activity to the EQ, lifting the air shelf by up to 3 dB as it works, so taming sibilance no longer dulls the top.
The controls
FREQUENCY
Where the detector listens for sibilance. 5 to 7 kHz suits most voices.
DETECT
HIPASS listens to everything above the frequency. BAND targets a single ess region for surgical work.
SENSITIVITY
Biases the adaptive threshold. Positive catches more esses, negative fewer. There are no attack or release knobs: it only reacts to sibilance, never to loud dark vowels.
RANGE
The most an ess can be turned down. Loud esses stay proportionally louder than quiet ones, so nothing sounds lisped.
MODE
WIDE turns the whole signal down during an ess. SPLIT dips only the sibilant band and leaves the body alone.
AIR LINK
How much de-esser activity lifts the EQ's air shelf, up to +3 dB. Taming esses no longer dulls the top.
06 · EQ
Twelve bands, surgical or broad.
A twelve-band equaliser with ten shapes, cut slopes to 96 dB per octave and dynamics on any band.
What it does
Bells, shelves, cuts, notches, tilt and all-pass shapes, each with its own power, solo and dynamic mode. Dynamic bands compress or expand only their own region, with a band-limited detector and an anti-pump release. Click on the curve to create a band where you point. Scale drags every band's gain at once, from flat to double, so a preset can be dialled back without redrawing it.
Why it's better
The filter mathematics is calibrated so the drawn curve is the curve you hear, with the frequency response verified against a professional reference equaliser. Analog Phase mode matches the phase of the analogue curve as well as its magnitude, with no pre-ringing and zero latency. Any structural change, a band switched on, a shape changed, a slope moved, crossfades in ten milliseconds, so the EQ can be adjusted while the singer is performing.
The controls
BANDS
Frequency, gain and Q per band, ten shapes, cut slopes from 6 to 96 dB per octave. Double-click a band to switch it off; solo auditions just that band.
DYNAMIC
Per-band dynamics: the band only moves when its own region gets loud, compressing or expanding by the range you set.
SCALE
Scales every band's gain at once: 100% as drawn, 0% flat, 200% double.
PROCESSING
Zero Latency matches the analogue magnitude curve. Analog Phase also matches the analogue phase, with no pre-ringing.
ANALYZER
Spectrum display: off, before the EQ, after it, or both, with a freeze for reading it.
07 · Catch
The fast one. Peaks caught, vocal kept forward.
A FET compressor in the classic fast style: a fixed threshold, an input you drive into it and ratios you punch in.
What it does
The detector sits after the gain stage, in a feedback loop, so it behaves like the hardware: more input means more compression, attack runs down to 20 microseconds, and release is programme-dependent, letting a sustained squash go more slowly than a brief peak. Ratios of 4, 8, 12 and 20 to 1, plus the all-buttons mode with its plateau curve and bite. Output is compensated automatically, so presets carry across levels.
Why it's better
The coloration follows the gain reduction the way a real FET stage does, odd-dominant and growing as you push it, instead of being a static saturation bolted on afterwards. The ratio buttons couple slope, threshold and knee together as one system, so each setting has its own character rather than being a number. Auto gain keeps the level honest while you find the sound.
The controls
INPUT
Drive into the FET compressor. The threshold is fixed like the hardware, so more input means more compression. Output is compensated automatically.
RATIO
4, 8, 12 or 20 to 1, or ALL: every button in, the aggressive all-buttons shape.
ATTACK
1 to 7, hardware style: 7 is the fastest, 20 microseconds.
RELEASE
1 to 7, 7 fastest. Programme dependent: a sustained squash lets go more slowly than a brief peak.
AUTO GAIN
Compensates the input drive so the level stays honest as you push it.
MIX
Parallel blend of compressed and dry.
08 · Level
Slow, musical, invisible.
An optical leveller after the catch, modelled on the classic two-knob studio compressor.
What it does
Peak Reduction sets how much the performance is levelled. A light-dependent cell model gives programme-dependent attack and a two-stage release that recovers half quickly and the rest slowly, following the last several seconds of the performance. Compress is a gentle 3 to 1; Limit is near-hard for catching peaks. Emphasis makes the detector more sensitive to brightness, so harsh moments trigger levelling sooner. Tube colour sits outside the loop, level-tracked.
Why it's better
This is the stage that keeps the vocal in the mix without anyone hearing it work. The release memory means a long phrase and a short shout are treated differently, the way the original hardware does it, and consonants pass through untouched. With the fast compressor catching peaks in front of it, Level only ever has to do the gentle part.
The controls
PEAK REDUCTION
How much the opto compressor levels the performance. Slow, smooth, and it stays out of the way of consonants.
MODE
COMPRESS is gentle, about 3 to 1. LIMIT is near-hard for catching peaks.
EMPHASIS
Makes the detector more sensitive to highs, so brightness triggers levelling sooner.
GAIN
Make-up gain after levelling.
AUTO GAIN
Compensates the levelling so loudness stays where it was.
