Effects

Reverb

Stereo reverb with both an algorithmic Freeverb engine and partitioned-FFT impulse response convolution.

What is a Reverb?

A reverb simulates the way sound behaves in physical space. When you clap in a cathedral, the sound bounces off walls, ceilings, and floors thousands of times before fading to silence. Each reflection arrives at your ears milliseconds after the original, slightly attenuated and slightly filtered. Reverb is the dense, evolving cloud of those reflections - what makes a recording sound *placed* in a room rather than floating in a vacuum.

Mathematically, a real space's response to a sharp click is its impulse response (IR). If you record that response, you can convolve any audio with it and the audio will sound like it was played in that space. Modern reverbs split into two camps: algorithmic reverbs that synthesize the response from networks of delay lines and filters (cheap, flexible, often more musical than literal), and convolution reverbs that use a recorded IR (literal, expensive, perfect for replicating a specific space).

The defining algorithmic architecture is the Schroeder reverb (Manfred Schroeder, 1962): parallel comb filters to build up density quickly, followed by allpass filters to smooth out the texture without changing the spectrum. Jezar's Freeverb (1999) refined this into a famously musical free-software reverb that became the basis for countless plugins. The Lexicon 224 and the EMT 250 are the hardware ancestors of every studio reverb sound you've heard since the 70s. Plate reverbs (the EMT 140) used vibrating sheets of metal; spring reverbs used coiled wire; both have wildly different characters.

The three knobs that matter: decay (how long the tail lasts), damping (how fast the highs fade in the tail - models air absorption and soft surfaces), and mix (how much wet signal vs. dry). After that it's algorithm-specific - room size, pre-delay, modulation depth, diffusion, early-reflection patterns. Reverb is the deepest effect on the chain.

Our Reverb

Our reverb is dual-mode: an algorithmic engine plus an impulse-response convolution engine, sharing one panel. Pick 'None' from the IR dropdown and you get the algorithmic reverb - a Freeverb-style network of 8 parallel lowpass-feedback comb filters into 4 series allpass filters. Decay sets the tail length, damping rolls off highs over time (modeling air absorption), size scales the comb delay lengths to change the apparent room dimensions.

Pick any other IR and you switch to convolution mode: uniform-partitioned overlap-save FFT convolution with 512-sample partitions and a 1024-point radix-2 FFT. This is the same technique used in commercial convolution plugins (Altiverb, Waves IR1) - low latency for live use, efficient enough to handle multi-second tails. Float64 internal precision avoids the rounding artifacts that show up in long convolution tails. In IR mode, decay becomes wet level and damping becomes a one-pole lowpass on the wet tail for tone shaping.

True stereo (L+R inputs and outputs - convolutions are per-channel), CV inputs on decay and damping, 10 HP. Place at the end of your chain, before output. For the kitchen-sink ambient reverb with shimmer and freeze, see Distant Horizons. For tape-flavored echoes-into-spring, see Secunda.

In a patch

Reverb almost always lives at the end of the signal chain, just before the output. The placement matters: if you reverb *before* a distortion, the distortion saturates the wet tail along with the dry signal, which sounds awful. Distortion first, then reverb, is the rule.

Mix to taste. 30-40% wet for a natural room feeling, 70-90% for ambient washes, 10-20% for subtle depth that doesn't push the source backward in the mix. Send-effect-style: 100% wet, with the dry signal going around the reverb on a separate mixer channel - this lets you fade the wet tail independently of the dry, which is how studio engineers use reverbs.

For pads and ambient, layer reverb after delay: the delay creates rhythmic repeats, the reverb smears them into a continuous wash. For shimmer, look at Distant Horizons, which adds octave-up pitch shifting in the feedback path. For granular ambient texture, pair with Clouds.

Inputs

IN L (audio) — Left audio input. If only IN L is patched, the signal feeds both channels for mono-to-stereo reverb.
IN R (audio) — Right audio input. Patch both for true stereo reverb processing.
Decay CV (cv) — Voltage control over the Decay parameter. Use an envelope to shorten the tail on percussive hits or an LFO for evolving room size.
Damp CV (cv) — Voltage control over the Damping parameter. Modulate this to change the brightness of the reverb tail over time.

Outputs

OUT L (audio) — Left output with dry/wet mix applied. Connect to Output or further processing.
OUT R (audio) — Right output. The stereo spread comes from offset delay lengths between left and right channels.

Controls

Decay — Reverb tail length (0.1 to 0.98). The knob is mapped to the audibly useful range of the underlying Freeverb network, so even low settings give a usable hall-sized tail. Around 0.4-0.6 reads as a medium hall, 0.7-0.85 as a long hall, and 0.9+ rings for many seconds (near-infinite wash).
Damping — High-frequency absorption (0 to 1). At 0, the reverb tail stays bright. As you increase damping, the highs fade faster than the lows, creating a warmer, more natural room sound.
Size — Room size (0.3 to 1). Scales all internal delay lengths. Smaller values sound like a closet or bathroom. Larger values sound like a concert hall. Changes the spacing and density of reflections.
Mix — Dry/wet blend (0 = fully dry, 1 = fully wet). Start around 0.3 for natural room reverb. Use 100% wet if patching as a send effect.

Inspired by

The algorithmic mode is a Freeverb-style reverb (Jezar's 1999 design): eight parallel lowpass-feedback comb filters into four series allpass filters, the Schroeder/Moorer formula that powers most software reverbs of the era. The IR mode is uniform-partitioned overlap-save convolution with a radix-2 FFT, the same technique used in Altiverb and Waves IR1, capable of multi-second tails at low CPU.

Freeverb (Jezar at Dreampoint)
Schroeder reverb topology
classic plate and hall algorithms

← Back to all modules