Creative Effects

VOID:SPACE

An original algorithmic reverb in the spirit of 1970s–1990s digital hardware, designed from first principles.

£39 £49Launch price

See the measurements

Formats: VST3 · AU · Standalone
Operating system: macOS · Universal
Version: v1.0.0

What it does

Ten spaces, measured into existence.

Lush, dense, smooth, musical and slightly coloured — built out of published DSP technique rather than reverse-engineered from anything. Early reflections feed a sixteen-line feedback delay network with an all-pass inside every loop, damping filters, modulation, and a level-normalised soft clip that sits inside the feedback path.

A mode is not thirteen knob positions. It also sets the tank's shortest and longest delay, so the spread is structural: a plate is short and tightly packed, a cathedral spreads across tens of metres. Size scales the whole set from there.

Wet level varies only 5.9 dB across all ten modes, so switching between them is not a volume jump.

Key features

Ten modes with structural differences

Room, Chamber, Bright Plate, Dark Plate, Hall, Concert Hall, Large Hall, Cathedral, Vintage Digital and a seeded Random Space. Measured ITDG runs from 3.8 ms on the plates to 46.5 ms on the cathedral; mid-band decay from 0.48 s to 6.30 s.

A decay time you can trust

In a parallel comb bank the slowest line dominates the late tail, so measured decay is set by whichever line happens to ring longest — 13.7 % high with gains nominally matched, 18.5 % with them fanned. Cross-coupling averages every line into one common rate: 0.7 % and 2.7 %.

In-loop colour that defaults to a true bypass

A cubic soft clip sits between the in-loop all-pass and the damping filter. Colour defaults to zero on every mode and zero is verified bit-identical. Modes set only a sensitivity, so a plate takes more grit than a cathedral at the same knob position.

Normalisation sized for a reverb, not a delay

A delay's repeats fall 20 dB; a reverb tail falls 60. The original 20× clamp bound after two seconds of a six-second decay, after which the normaliser was inert. Raised to 256× with an asymmetric envelope, drive reaching the curve at three seconds is now 35 dB higher than a plain in-loop saturator.

Anti-aliasing that was measured, then removed

The rest of the range antialiases its nonlinearities and this one started that way. At the drive this tank actually reaches, the pointwise curve aliases at −145 dB and ADAA made it 41 dB worse. Removing it also restored a pointwise stability proof in place of an empirical one.

A 5.8× speedup from 64 bytes

The tank's sixteen delay lines cost 62 ns per sample for nothing but sixteen reads and writes. The cause was cache set aliasing — a power-of-two capacity meant every line shared its low address bits. Padding the stride by a single cache line took it to 10.79 ns.

Audio

Audio demos for VOID:SPACE are not published yet.

We would rather show you nothing than show you a demo that was not recorded through this plugin.

Measured

Stage test benches · Apple Silicon · 48 kHz

Whole plugin, Colour off: 121 ns/sample · 0.58 % of one core
Whole plugin, Colour on: 156 ns/sample · 0.75 %
Decay accuracy, mid-band T60: within 1.3 %
Tank delay lines, before and after stride padding: 62.44 → 10.79 ns
Wet level spread across all ten modes: 5.9 dB
Early reflection spread across input correlation: 0.01 dB
Mode-switch discontinuity once the tank has flushed: 0.0016
auval: AU VALIDATION SUCCEEDED

Figures are produced by this plugin’s own test suite and benchmark, on the hardware and sample rate named above. They are not estimates.

Technical details

Formats: VST3 · AU · Standalone
Operating system: macOS 10.13 or later
Architecture: Universal — Apple Silicon and Intel
Plugin code: Sgvb
Version: 1.0.0
CPU: 0.58 % of one core at 48 kHz, 0.75 % with Colour engaged

Worth knowing

The plugin binary ships under the product name VOID SPACE. The colon is a brand mark used on this site.
The full control surface is not yet documented in the repository; the controls section will be published once it is.