Skip to main content

​
ANTIVERB: Psychoacoustic Densifier

​
How It Works

ANTIVERB runs a source through six processing stages (Source, Densification, Early reflections, Diffusion, Late tail and Tone), each with its own MIX and BYPASS. Every stage MIX starts at zero, so a fresh instance passes the source unchanged. Raise each stage’s MIX to add that stage. Source. Warmth, Brightness and Presence shape the incoming signal before the later stages hear it, and Distance and Air Absorption place the source 0.5 to 20 metres from the listener, on the dry signal and on everything it feeds into the reverb, with an air-absorption model (ISO 9613-1) that takes high frequencies away as the distance grows. Densification. Densification runs the source through a cascade of sparse FIR filters whose taps add up to between 10,000 and 200,000 effective reflections at the default depth. Where those taps fall in time is set by one of eight placement profiles (Regular, Golden, Even, Natural, Branching, Prime, Chaos and Voice), and each profile gives the same source a different grain. Depth splits the tap count across one, two or three cascaded stages, trading per-stage character against smoothness and CPU. A three-part envelope (drop, hold, decay) shapes the field over a length of up to 8 seconds, and Punch lifts or softens the leading edge. Sheen sets how much of the cascade returns beside the source and stops at 50 %, the point past which a cascade this dense starts to read as comb filtering. A resonance smoother follows the cascade and cuts only the peaks that rise above the local spectral trend, and a bass split gives the low band its own, sparser filter so the bottom stays articulate under a dense top. Early reflections. Early reflections arrive in three groups (near, mid and late), each with its own level and time window. Diffusion. A chain of up to twelve nested allpass stages blurs the reflections into a diffuse cloud on their way to the late tail. Late tail. A feedback delay network of 16 to 128 lines carries the tail, with separate decay times for bass, mid and treble between 20 milliseconds and 12 seconds. The network’s delay lengths follow whichever placement profile is active, so a change of profile changes the tail along with the densified field. Tone. After every reverb stage, a single Tilt control leans the whole output brighter or darker across 31 bands from 20 Hz to 20 kHz. Spatial. On a stereo track ANTIVERB is a stereo processor. On a multichannel track each part of the sound becomes its own object in the speaker field: Dry, Density, Reflections and Space each get an azimuth, an elevation, a width and a spread, and each can be placed independently. Every tap, reflection and delay line is panned across the rig with VBAP and MDAP spread,1 so the texture of the densifier is distributed over the speakers tap by tap. The rig is drawn around the listener and can be any of the standard beds (stereo, quad, 5.1, 7.1, 5.0.4, 7.1.4, 9.1.6) or a custom arrangement of up to 16 speakers placed by X, Y and Z, with per-speaker time alignment and bass management.

​
Key Capabilities

  • Sparse-FIR densifier in a one-, two- or three-stage cascade, with 10,000 to 200,000 effective taps in the two- and three-stage settings
  • Eight tap-placement profiles: Regular, Golden, Even, Natural, Branching, Prime, Chaos, Voice
  • Feedback delay network of 16 to 128 lines with bass, mid and treble decay from 20 ms to 12 s
  • Early reflections and diffusion: three timed reflection groups and up to twelve allpass stages
  • Resonance smoother that cuts peaks above the local spectral trend inside a chosen frequency range
  • Per-stage spatial objects panned with VBAP and MDAP spread
  • Speaker rigs up to 16 outputs, stereo through 9.1.6 or a custom layout, with time alignment and bass management
  • MINIMAL and EXPERT views and an in-plugin Codex
  • User presets, with an A/B morph driven from host automation
  • VST3 and Audio Unit on macOS 13.0 or later (Apple Silicon and Intel); VST3 and a standalone application on Windows 10 or later (64-bit)

Footnotes

  1. V. Pulkki, “Virtual Sound Source Positioning Using Vector Base Amplitude Panning”, Journal of the Audio Engineering Society 45(6), 1997; V. Pulkki, “Uniform spreading of amplitude panned virtual sources”, Proc. IEEE WASPAA, 1999. ↩