NOMN

Reverb and reverberation is a uniquely special audio effect and a contested area in acoustics design. The underlying goal is to give acoustic sound (either in a real space or from a speaker) a type of presence or perceptual gravity. This has an architectural and archeo-acoustic history of its own, and the desire to feel enveloped by and have the sound we make extended into space larger than ourselves is perhaps as old an acoustic desire as any.

In architecuture, this is usually handled with standards for how particular spaces should reverberate or not. We measure decay times, calculate absorption coefficients, ray trace model design documents, but reverberation is only one layer of how a sound's presence in a room takes shape. It doesn't tell the whole story, it's a general range, a shorthand.

Reverb plugins tend to point toward placing audio in a virtual space, technically and poetically. A preset might invoke a fantastical space, even without air (outer space) or a space with totally different acoustic parameters and sound speed (underwater), model a specific famous room with an impulse response, or contain a unique approach to algorithm creating a sense of room-ness. In the end, the metaphor tends to be about relocating audio from one non-space (computation) into some "real" space, yet this too is also "just" computed.

Convolution reverbs sample real rooms and replay their impulse responses (your sound through another sound). Algorithmic reverbs answer it by simulating idealized rooms, using feedback delay networks to approximate the late tail of a hall and plate models to approximate the resonance of a steel sheet. Digital reverb engineering has grown increasingly accurate at answering this single question, and there is very little reason to reinvent any of the amazing reverbs out there.

But if the goal of this kind of computation is something broader and more particular than a virtual space, what other approach could we take?

An adjacent question to all this that's a bit stranger: what makes the brain perceive a sound as full and embedded psychically when no space has been modeled at all?

Most of the audible character a listener associates with reverb comes from a single sound being multiplied into many time-staggered copies, dense enough that the brain stops resolving them individually as echoic copies and starts hearing a thicker, richer audio "thing". Real rooms produce this through reflections flying around creating compressive near-mass at the speed of sound, a form invisible to the eye but sensable through the ears and skin.

But what if there were compound perceptual effects in this territory that need no room as its model, nor a realistic sense of scale or pattern- just high density time-staggered copies arriving at significant, patterned density with a particular aesthetic/temporal structure that seems to engage the perceptual system that maps sound to space?

ANTIVERB makes this spculation about psychacoustic density the first stage in a multistage presence processor that allows a user to layer more traditional algorithmic reverb with this "antiverb". For this reason we call it a psychoacoustic densifier in that it creates a marked effect on this layer of how we perceive a recording, accepting it as a colorization or enhancement, not an attempt to recreate anything.

ANTIVERB is a chain of six stages that run in series, each one processing the output of the stage before it. Every stage has its own wet/dry mix, bypass and solo, so the sound can be built up one layer at a time and any single layer can be heard on its own or used on its own. Input and output trims sit at either end of the chain, and a global dry/wet sets the overall balance.

1. Source. Before anything is added to it, the dry signal can be set at a distance from the listener, with high-frequency air absorption scaled to that distance, and shaped with warmth, brilliance and presence controls.

2. Densification. The core engine is a sparse-FIR cascade: a structured array of thousands to hundreds of thousands of carefully-placed reflections, computed at every audio sample. The placement of those reflections is what determines the character of the densification. ANTIVERB ships with eight tap-placement profiles, each derived from a different generative law: golden-ratio quasiperiodic, Fibonacci spiral, prime-number sieve, even spacing, deterministic chaos, voice-cluster patterns, recursive branching, and a baseline regular distribution. Each profile produces a sonically distinct signature that the user can audition without dialing. The cascade runs one, two or three generators deep, each multiplying the effective density of the one before. Tails run from a few milliseconds up to 8 seconds, shaped by a three-part envelope of rapid initial drop, flat plateau, and a linear, concave or convex release. A two-band mode splits the input at a crossover and sends the low end through a sparser tap pattern so the bass keeps its articulation, and optional 2× oversampling reduces aliasing in the high-frequency fill. A psychoacoustic resonance smoother (the "tamer") suppresses peaks that protrude above the local spectral trend without flattening the overall response, working on left/right or mid/side with an adjustable tilt target.

3. Early reflections. Three groups of discrete reflections (very early, a cluster, and a late cluster), each with its own level and end time, give the densified sound its first sense of surrounding surfaces.

4. Diffusion. A short allpass network, coupled between mid and side, smears the reflections into a continuous wash before the tail, with adjustable stage count and maximum delay.

5. Late tail. A feedback delay network of 16, 32 or 64 lines handles the late tail, with a choice of Householder, scattering or circulant feedback matrix (plus an optimized colorless circulant at 16 and 32 lines for a flatter tail), separate decay times for low, mid and high bands, pre-delay, input diffusion, and sine or noise modulation. The FDN's per-line delay times inherit the active densifier profile's centroid, so the tail stays coherent with the densification it follows.

6. Tone. A tilt EQ (±6 dB) sets the tonal balance of the whole output.

The densifier, reflections and tail route directly into the output bus, including multichannel buses of up to 16 channels in any arrangement. ANTIVERB's spatial routing is per-tap: each individual reflection is panned across the active speaker layout via Vector Base Amplitude Panning (VBAP), so the same algorithmic densification automatically distributes itself across stereo, 5.1, 7.1.4, or any user-defined irregular speaker rig. The spatial signature is derived from the same profile that shapes the temporal signature.

Field controls rotate, tilt and spread the whole field and weight it toward the front or back, and the densifier, reflections and tail can each be taken out of the spatial routing on their own. A speaker layout editor places the speakers and the listener in a 3D view, saves and recalls custom layouts, and time-aligns each speaker so that all of them arrive at the listener together; a coverage heatmap shows how evenly the wet signal covers the room floor. Bass management sends everything below an adjustable crossover (40 to 200 Hz) to the subwoofer at its own level, and each output can be given a role and a trim.

Seven macro knobs sit across the tabs (Shape and Width on densification, Reflections and Room on the reflections and diffusion, Decay Color and Motion on the tail, and Width on the spatial field), each moving a group of related parameters at once. Two saved presets can be loaded into A and B slots and morphed continuously, and the morph position is automatable, so a mix can travel from one setting to another over the course of a song. A Codex panel slides over any tab and explains every control on it, including what each one does to the sound.

ANTIVERB can produce sounds that resemble rooms when configured to do so, and its native register goes where rooms cannot: a sustained note thickens while its tail stays the same length, a vocal gains presence and air with no audible space forming around it, a drum room picks up body without getting longer, and an immersive bed envelops an object without placing it at any specific location.

Vocals. Densification at short decays produces the fullness classically associated with high-end console reverb sends, and the vocalist stays in the room they were recorded in. It works particularly well on close-mic'd vocals, where the dry signal already has its own intimacy.

Drums and percussion. Short densifier tails behave as transient extenders, thickening the attack while the source keeps its apparent space. Longer tails add the sustain a room would give the kit, and the kit's own recorded space stays as it was.

Mixing and mastering. ANTIVERB engages cleanly at low intensity values where the signal flow is largely transparent, producing a thickness that can be added to a master without committing to a sense of space.

Immersive (5.1, 5.0.4, 7.1.4, 9.1.6, custom). Per-tap VBAP routing distributes the densification across whatever speaker layout the host bus advertises, including arbitrary irregular rigs.

The signal path contains no impulse responses and no samples of any room. The algorithm has no model of physical acoustics (walls, absorption coefficients, ray tracing), and it emulates no plate, spring or hardware unit.

Mac: VST3 and AU, universal (Apple Silicon and Intel), macOS 13.0 or later. Windows: VST3 and Standalone, Windows 10 or later (64-bit). Stereo input. Output: stereo up to 16 channels in any arrangement. CPU footprint scales with density and channel count; defaults run comfortably on modest hardware.

Includes a 14-day free trial with full functionality and no watermark or save restrictions. After the trial, audio mutes until activation.