Ten character plugins — now live

Ten characters,
derived from research.

Most plugins are guessed. Ours are derived — every coefficient traces to a peer-reviewed paper. Schlecht & Habets on FDNs. Fagerström on dark velvet noise. Huovilainen on analog filters. Arnardóttir & Smith on tape echo. Pakarinen & Yeh on tube saturation. Schroeder & Dattorro on long allpass cascades. Linn-firmware-derived swing curves on the Groove Engine. Kac on the spectrum of a planar domain. Das, Schlecht & Välimäki on coupled-volume acoustics. Ten plugins, ten distinct moods, one coherent catalog.

Scatter is the room you stepped into. Wash is the reverb already in the room with you. Grain is the song made on hardware. Echo is the sound bouncing back at you. Tape is the master bus the song lives inside. Bloom is the warmth the sound grows. Drift is what happens when the room runs out of walls. Crate is the kit the song was sampled from. Manifold is the room solved from an equation, not tuned by ear. Spaces is the room next to the room.

Mac · Windows · Linux CLAP · VST3 pip install clixhouse

SCATTER
REVERB · v0.1
SIZE
42%
DECAY
1.4 s
DAMP
55%
MIX
40%
OUT
−5.4 dB
FDN16 · 48 kHz · 0.34% CPU Schlecht-Habets · Jot · Huovilainen

Visual mockup. Actual UI in development.

Hear it

Same source. Different rooms.

Hit play. Switch presets. These are real renders of synthesized source through the actual Scatter engine — no smoke, no mirrors.

NOW PLAYING
— pick a clip below —
0:00 / 0:00
Vocal a cappella · the snake-charmer test
Real drum loop · 90 BPM, captured kit
Vocal melody scale · long sustained notes
Piano melody · slow, expressive phrasing
Acoustic guitar · live arpeggio
Synth lead · sustained looped melody
01 Boom-bap drum loop
02 Pentatonic pluck melody
03 Sub bass groove
04 Slow pad progression
05 Snare hit · the canonical reverb test

Real-instrument CC0 sources from Freesound at the top. Nine named presets, each a deliberate arrangement of the engine — not a parameter dump. Hover any preset to see what it's built for. Synthesized clips below are engineering reference signals.

A second reverb — Wash

Sense of space vs sense of time.

Scatter is the room you stepped into. Wash is the reverb already in the room with you — the warm patina of tape, the grit of vintage hardware, the audible character of analog noise. Built on Dark Velvet Noise (Fagerström et al. 2024 JAES), it's textured where Scatter is smooth. Different feelings. Most producers who care about reverb end up wanting both.

Vocal a cappella · the same vocal, the other reverb
Drum loop · texture without losing the punch
Piano · sparse pulses or smoke, your call

Seven Wash presets, each a sparse-rectangular-pulse FIR arrangement — not Scatter at lower mix. Full character map + design philosophy in paper/WASH-PRESETS.md.

A third product — Grain

Makes new sounds sound old.

Scatter is the room you stepped into. Wash is the reverb already in the room with you. Grain is the song made on hardware — a fused SP-1200 quantizer + SSM2044 ladder filter signal chain. The character processor producers reach for when they want a cleanly-recorded source to feel like it was sampled into an SP-1200 in 1989. Different from a reverb in every measurable way; sits next to Scatter and Wash on the channel strip without competing.

Vocal a cappella · the same voice, through the era
Drum loop · the canonical SP-1200 break, or further
Dirty drums · grittier source, same engine — different floor
Synth lead · where the resonant filter presets actually sing
Piano · sampler character on top of clean piano

Eight Grain presets across the (bit depth × sample rate × cutoff × resonance × drive × pitch) space. Full producer manual + engine rationale in paper/GRAIN-PRESETS.md.

A fourth product — Echo

Sound bouncing back at you.

Scatter is the room. Wash is the reverb already with you. Grain is the song made on hardware. Echo is the sound bouncing back at you — a tape echo modeled on the Roland RE-201 family. Multi-head delay topology with wow-and-flutter modulation per head, soft-knee tape saturation, tape-age aging. Different category entirely from the reverbs and the lo-fi chain; sits in your DAW's Delay folder, not the reverb folder.

Vocal a cappella · the same voice, bouncing back
Drum loop · slap, multitap, or self-oscillating mayhem
Piano · rhythmic delay, multitap movement, ambient wash

Eight Echo presets — multi-head tape delay topology with wow, flutter, saturation, and tape-age aging. Full producer manual + engine rationale in paper/ECHO-PRESETS.md.

A fifth product — Tape

Tape is the master bus the song lives inside.

Tape is the master bus the song lives inside — a transport simulation, not an insert effect. Input transformer (solid-state, tube, or coupled), bias-modulated compression, bus-scale wow and flutter, head bump, speed-dependent HF rolloff, cross-channel low-bleed. Three stock voicings: 2-inch master, 1/2-inch, cassette. Drop it on the 2-bus, walk away, and the mix stops sounding like it was assembled in a DAW and starts sounding like it lives somewhere. Lives in your DAW's Mastering folder, not next to the reverbs.

Full mix · the entire song through the 2-bus
Vocal · the same voice, printed to tape
Drums · pushed into tape, ranging from warm to wrecked
Piano · the same notes, on five different machines

Eight Tape presets across the (input × bias × transformer × stock × tape speed × wow × head bump × highs × crosstalk) space. Full producer manual + engine rationale in paper/TAPE-PRESETS.md.

A sixth product — Bloom

Bloom is the warmth the sound grows.

Bloom is a harmonic saturator — the channel-strip effect producers reach for on almost every track without thinking. Vocals to cut through without an EQ boost, synths to feel alive, drums to grow edges, bass to stop sounding like a sine wave. Three voicings each emphasize a different harmonic profile (Tube with gain-asymmetric clipping for 2nd-harmonic warmth, Tape with pure tanh for smooth 3rd-harmonic body, Transformer with arctan for elevated 5th-harmonic snap). A presence parameter blends along the 2nd/3rd/5th axis — not a brightness slider; the implementation actually shifts which harmonics dominate. Single-band or three-band drive via Linkwitz-Riley crossovers. 4× polyphase oversampling around the shape block. The bar: a producer drops Bloom on a thin vocal, walks away, and the vocal sits in the mix without anyone needing to ask which knob did it. You hear it as absence when you bypass it. Lives in your DAW's Distortion folder alongside Grain — Grain is era-specific style you HEAR, Bloom is universal warmth you hear as ABSENCE.

Vocal · the body that wasn't there in the DI
Vocal scale · two voicings on the same lines
Bass · the sine-bass complaint solved
Sine bass · the literal sine bass that hears Bloom as body
Pad · DAW-flat to breathing
Plucked melody · life vs bite
Drums · transient bite to wrecked
Snare · crack from edge to wrecked

Eight Bloom presets across the (voicing × mode × drive × presence × bias × per-band drives × crossovers × makeup × mix) space. Full producer manual + engine rationale in paper/BLOOM-PRESETS.md.

A seventh product — Drift

Drift is what happens when the room runs out of walls.

Drift is an infinite-sustain ambient reverb — the channel-strip effect producers reach for when a sustained chord needs to transcend itself. Pads that hold for sixty seconds without decaying, single notes that become weather, ambient passages where the reverb IS the song's third instrument. The engine is a 32-stage Schroeder allpass cascade per channel (long cascades produce smooth diffusion without the modal ringing that 4–8 stage cascades leave audible at long decays) with per-stage decorrelated LFO modulation (essential, not optional — static long cascades comb-filter), a pitch-shifted feedback path for the octave-up shimmer signature (granular two-pointer pitch shifter, default +12 semitones), and a three-band feedback EQ (200 Hz / 1 kHz / 4 kHz) that controls which frequencies sustain longer than others. No early reflections by design — Drift is not about a space, so the cascade's first stage IS the pre-delay if presets want a longer onset. The bar: a producer holds a major-7 pad for four bars, walks away, comes back to a room transformed — the sustained chord didn't fade, it became the texture the song lives inside. Lives in your DAW's Reverb folder alongside Scatter and Wash — Scatter is the clean spatial room, Wash is the textured nostalgic patina, Drift is what happens after space gives up. Wash makes a sound feel REMEMBERED; Drift makes it feel INHUMAN.

Pad chords · four configurations of held sustain
Vocal · ethereal to brooding
Vocal scale · cathedral vs endless
Piano · held resonance painted long
Plucked melody · three flavors of shimmer
Guitar arpeggio · ambient bed under fingerpicking

Eight Drift presets across the (size × decay × shimmer × modulation × feedback-EQ × mix) space. Full producer manual + engine rationale in paper/DRIFT-PRESETS.md.

An eighth product — Crate

Crate is the kit the song was sampled from.

Crate is a 16-pad sampler instrument — the first non-channel-strip in the catalog. Where Scatter, Wash, Grain, Echo, Tape, Bloom, and Drift live on a track, Crate is the track. Load samples onto pads, sequence patterns, hear them through SP-1200 quantization and SSM2044 ladder filtering per pad, with Crate Glue (summing nonlinearity + SSL G-style bus comp), inline Bloom (the catalog's harmonic saturator), and inline Tape (the catalog's master-bus transport) on the bus. The differentiator is the Groove Engine: nine swing templates derived from primary sources — Roger Linn's documented MPC60 and MPC2000 firmware curves, GMD humanization aggregate stats, and placeholders for the producer-specific templates (Dilla Time, Premier, Pete Rock, Madlib Drift, Just Blaze Push) pending Cobb-style canonical-record transcription. Sample-accurate microtiming per step, not block-quantized. The bar: a producer programs a snare on steps 0, 4, 8, 12 at 92 BPM, switches the groove from Straight Grid to MPC60, and immediately hears the 58% swing delay-the-and feel that Linn put in the original firmware in 1986. Ships as CLAP + VST3 (Instrument / Drum category), with a headless CLI binary (build/crate) and Python bindings (pip install clixhouse) for AI-driven beat generation. Bit-exact parity gate: the same kit + pattern renders byte-for-byte identical through the Python ABI, the CLI binary, and the CLAP plugin host.

Try the 4-pad demo in your browser →

See it before you hear it — the same 16th-note hat row, three groove templates

step 0 4 8 12 16 STRAIGHT MPC60 MPC2000

MPC60 shifts the odd-step hats by ~10 ms (the "and" of each beat lands late) — Roger Linn's documented 58% swing curve. MPC2000 is lighter (~5 ms, 54% swing). Straight Grid is the reference. Press the buttons below to hear the difference on the same kit and the same pattern.

Same kit, same pattern · four grooves — hear the Linn-firmware swing

Four grooves through the same kit + same pattern at 92 BPM. The audible difference is the Groove Engine — sample-accurate microtiming per step, derived from Roger Linn's documented MPC firmware curves and the GMD humanization aggregate. v0 plugin ships the engine + bus chain + sequencer + groove engine. GUI (4×4 pad grid + sample browser + sequencer view) lands in v1. Factory preset kits — the producer-vetted Boom-Bap, Trap, House, Lo-Fi, Vinyl-Hip-Hop — ship alongside the GUI. Full spec + design rationale in docs/CRATE-SPEC.md.

A ninth product — Manifold

Can you hear the shape of a drum?

Manifold is a reverb whose modal density is a real eigenvalue problem. Where Scatter / Wash / Drift each start with a chosen reverb topology (FDN / Dark Velvet Noise / Schroeder cascade) and let the math fall out, Manifold goes the other direction — pick a shape, solve the 2-D Helmholtz equation on it, and the modal density falls out of physics. Six factory presets, each a different planar domain:

The math is real. The level-spacing statistics in every shipped preset are gated in CI on every push — we re-derive Wigner-Dyson GOE for the chaotic shapes, Poisson for the rectangle, near-equal spectra for the Twin Shapes pair. The audible novelty isn't a marketing claim — it's an invariant the code refuses to ship without.

Citations: Kac 1966, Bunimovich 1979, McDonald-Kaufman 1979, Bohigas-Giannoni-Schmit 1984, Heller 1984, Gordon-Webb-Wolpert 1992, Sridhar 1991, Penrose 1974, Jot 1991, Zavalishin 2012, Steiglitz 1996, Rössler 1976.

Try the Twin Shapes A/B demo in your browser →

The single strongest demo: drop the same source into two parallel tracks, insert Manifold on both, select Twin Shapes A on one and Twin Shapes B on the other. Identical knob positions. The IRs differ in spatial image and fine temporal structure. The static spectrum metrics agree to within 5%. You're hearing the shape of the drum that Kac proved you couldn't hear from the spectrum alone. Full citations + math derivation in paper/MANIFOLD-DESIGN.md; preset manual in paper/MANIFOLD-PRESETS.md.

A tenth product — Spaces

Most reverbs simulate one room. Spaces simulates connected rooms.

Spaces is a coupled-volume reverb built on a Grouped Feedback Delay Network. Where Scatter / Wash / Drift / Manifold each model a single space, Spaces models two or three connected ones: a cathedral nave opening through arches into side chapels, an apartment bleeding into a long hallway, three coupled caverns answering each other a beat late. The dry signal enters a source room; energy bleeds into the coupled volumes through frequency-dependent apertures — highs couple less than lows, the way a real doorway passes bass more easily than treble. The result is a decay that changes slope partway through: a fast initial tail from the room you're in, handing off to a slow late tail from the volume next door. No single-volume reverb can reproduce this multi-stage, non-exponential decay by tuning — the structure has to be there. Lives in your DAW's Reverb folder alongside Scatter, Wash, Drift, and Manifold — the fifth answer to a different question: Scatter is the room you stepped into; Manifold is the room solved from an equation; Spaces is the room next to the room.

The research basis is Das, Schlecht & Välimäki, "Grouped Feedback Delay Networks With Frequency-Dependent Coupling," IEEE/ACM TASLP 2023 — the first commercial plugin to ship coupled-volume acoustics. Eight factory scenes (Cathedral, Apartment, Grand Foyer, Stairwell, Caverns, Underpass, Basilica, Cloister), a preset selector plus five macros (Mix, Width, and scale knobs for Size, Decay, and Coupling). The Coupling control is decay-invariant by construction and the network is unconditionally stable at every setting — push it far and the far room sustains into a bounded ambient drone, never a runaway.

Try Spaces in your browser →

Snare · the slap crosses into the next room and slows
Plucked melody · dense coupling vs the hero cathedral
Pad chords · the far volume takes over the tail
Vocal · sacred and wide vs blooming late
Synth lead · the far chamber answers a beat late
Drums · hard-coupled mid-forward concrete

Eight coupled-volume scenes across the (per-room size × decay × damping × coupling) space. The signature is most obvious on a short transient — hear the near room decay, then the coupled volume swell up underneath and carry the tail. Full producer manual in paper/SPACES-PRESETS.md; coupled-volume design memo + citations in paper/grouped-fdn-design.md.

Ten products — not ten versions of one thing

Buy one, or buy the catalog.

Most producers who care about character end up wanting more than one. The ten characters sit next to each other on a channel strip, on the master bus, and on the instrument browser without competing — different DAW folders, different audit methodologies in the engineering, different roles in the song. Buy the catalog and the per-product price drops at every tier.

Any one

Single product

full price

Pick the one you came for. Scatter, Wash, Grain, Echo, Tape, Bloom, Drift, Crate, Manifold, or Spaces — each ships standalone with its full preset bank, both CLAP and VST3 (instrument or effect category as appropriate), every supported OS.

Reverb pair

Scatter + Wash

−20% together

The space-and-time pair. Scatter for clean spatial reverb, Wash for textured tail. Most producers who buy one come back for the other within a month.

Reverb trio

Scatter + Wash + Drift

−30% together

Three reverbs, three distinct moods. Scatter is the room you stepped into. Wash is the reverb already in the room with you. Drift is what happens when the room runs out of walls. Three different answers to three different questions.

Ambient pair

Wash + Drift

−20% together

For producers who live in ambient, post-rock, or film-score territory. Wash makes a sound feel REMEMBERED; Drift makes it feel INHUMAN. Two different sustained-tail flavors that compose more often than either composes with the other five.

Character pair

Grain + Bloom

−20% together

The two Distortion-category products. Grain is era-specific style you HEAR (SP-1200 in 1989); Bloom is universal warmth you hear as ABSENCE when you bypass it. Two different needs, one folder in your DAW.

Lo-fi trio

Wash + Grain + Echo

−30% together

The full lo-fi character chain. Wash for time, Grain for medium, Echo for repetition. Buy if you make hip-hop, vaporwave, ambient, or anything that wants the era to be heard.

Master bus essential

Tape standalone

full price

Tape on the 2-bus is what producers reach for when a rough mix coheres. Buy it standalone if it's the one CLIXHOUSE product that fits your workflow — the master-bus character pass, not a channel-strip insert.

The full catalog

All ten

−50% together

Scatter + Wash + Grain + Echo + Tape + Bloom + Drift + Crate + Manifold + Spaces. The catalog as one purchase, at the lowest per-product price we offer. Nine channel-strip and master-bus effects and the 16-pad sampler instrument — the full kit, from the kick drum you start with to the room you put it in. Every preset bank, every documented audit, every supported plugin format.

Pricing finalized at launch. The bundle ratios are firm; the dollar amounts depend on what producers tell us they're worth during beta. Join the waitlist below to lock in the launch-day pricing.

Downloads

Get the plugins.

The beta bundle ships every product as native CLAP + VST3. Linux and macOS (Apple Silicon) are available now; Intel-Mac and Windows builds are on the way. Mac builds aren't notarized yet — your DAW loads them after a one-time "allow" (instructions are in the download).

Linux x86_64

Available now

All ten products, CLAP + VST3, in one archive. Unzip into your plugin folder and rescan.

Download Linux bundle (v0.2.0) →

macOS (Apple Silicon)

Available now

All ten products, CLAP + VST3, for M1/M2/M3/M4 Macs. Not notarized yet — a one-time "allow" step is in the download. Intel-Mac build coming.

Download macOS bundle (v0.2.0) →

Windows

Coming soon

x86_64 build (CLAP + VST3) — compiling in CI now. Join the waitlist to hear when it lands.

Research-grade

We measure ourselves against real rooms.

Not a sales pitch — actual measurements. We ran Scatter through the same acoustic test battery we'd apply to a real recording of York Minster.

1.72 s
Scatter hall RT60
Real hall: 1.84 s · 6% match
0.34%
CPU per instance
AVX2 · 48 kHz · 256-sample block
< 1 ULP
Plugin vs Python parity
Same sound everywhere it runs
7
Memory-leak-free tests
Valgrind-clean, every destroy path

Built on Schlecht & Habets (2020) feedback delay networks · Jot (1991) absorption design · Huovilainen (2004) analog character modeling. Full measurement report on request.

For developers

The same reverb. From Python.

Scatter ships as a plugin and a Python library, bit-identical. Build AI music agents, batch-process audio for games, run reverb inside a Jupyter notebook.

# pip install clixhouse
from clixhouse import Reverb

reverb = Reverb(preset="4am_bedroom")
wet = reverb.process(dry_audio, sample_rate=48000)

No other audio company is shipping a serious DSP library for AI agents. We're betting that changes fast.

Be the first to hear it on your tracks.

No spam. One email when the beta opens, one when v1 ships. That's it.

Honest FAQ

Does it sound better than Valhalla?

Different. Scatter has a slower diffusion buildup than most plugin reverbs — closer to vintage algorithmic units than to convolution. Try the demos. If it doesn't fit your taste, Valhalla is great.

Why is the source material synthesized?

Licensing. Real stems from commercial sources can't be redistributed. The synth is honest about what the reverb does — your real vocals will hit harder.

When does the plugin ship?

The audio engine is production-ready today. The UI shell and Mac/Windows installer are in progress. We'll email the waitlist before we ship to the public — no exceptions.

What's the price?

Single-product pricing: $39 introductory. We'll publish a fair-pricing breakdown alongside launch.

Why "Scatter"?

Sound scatters in a room — that's what reverb is. We chose the physics word.